{
  "as_of": "2026-10-30",
  "models": [
    {
      "id": "claude-opus-5",
      "name": "Claude Opus 5",
      "vendor": "Anthropic",
      "released": "2026-07-23",
      "context_window": 1000000,
      "price_input": 5,
      "price_output": 25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": 93.9,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning",
        "long-context",
        "research"
      ],
      "strengths": [
        "Tops Anthropic's Frontier-Bench v0.1 agentic-coding run at 43.3% — above Claude Fable 5 (33.7%) and more than double Opus 4.8 (18.7%), at a lower cost per task",
        "Fable-class coding at half Fable's price ($5/$25 vs $10/$50) — on CursorBench 3.2 at max effort it lands within 0.5% of Fable 5's peak score at about half the cost per task",
        "Five-rung effort dial (low → max) so you can trade cost against capability per request; low/medium are unusually strong on this model",
        "1M-token context as both default and maximum, 128K max output, and a 512-token prompt-cache minimum (half Opus 4.8's)"
      ],
      "weaknesses": [
        "SWE-bench Verified isn't published yet, so the coding score is an agentic-suite-led estimate; GPQA (93.9) is Epoch's independent run at stated effort — best-case, not default",
        "The headline Frontier-Bench comparison is Anthropic's own internal run on its own harness — directional for cross-lab ranking, not settled",
        "Still Opus-priced: $5/$25 is far above Sonnet 5 or the open tiers for bulk work",
        "Thinking is now on by default, and disabling it is only allowed at 'high' effort or below — quiet cost and behaviour change if you're porting Opus 4.8 code across"
      ],
      "verdict": "The new default for hard coding and agent work: it matches or beats Fable 5 on the agentic suites its own lab publishes, at half the price — just treat the benchmark picture as early until third parties verify it.",
      "confidence": "medium",
      "sources": [
        "https://www.anthropic.com/news/claude-opus-5",
        "https://platform.claude.com/docs/en/about-claude/pricing",
        "https://www.cnbc.com/2026/07/24/anthropic-claude-opus-5-ai-fable-5-cost.html",
        "https://venturebeat.com/orchestration/anthropic-launches-claude-opus-5-a-cheaper-ai-model-for-coding-agents-and-enterprise-workflows",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/document",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/vision",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": 95,
      "coding_basis": "Frontier-Bench v0.1 43.3% (#1, ahead of Fable 5's 33.7%) and CursorBench 3.2 within 0.5% of Fable 5's peak at max effort — calibrated against Fable 5 (95). SWE-bench Verified not yet published, so this is an agentic-suite-led estimate, not a SWE-bench number.",
      "coding_confidence": "medium",
      "use_well": [
        "Start at 'xhigh' effort for coding and agentic work, 'high' elsewhere — then sweep down. Low and medium punch well above their weight here, and effort is the main cost lever.",
        "Give it the whole task spec in one first message and let it run — it's built for long autonomous sessions, and drip-feeding the task across turns costs more and works less well.",
        "Thinking is ON by default now (it wasn't on Opus 4.8) — so re-check your max_tokens, or responses can truncate mid-answer.",
        "It verifies its own work without being asked. Delete any 'double-check your answer' instructions you carried over — they now cause redundant work."
      ],
      "price_checked": {
        "url": "https://platform.claude.com/docs/en/about-claude/pricing",
        "date": "2026-07-29",
        "input": 5,
        "output": 25,
        "aggregator_note": "OpenRouter shows ~$50 — that's Fast mode ($10/$50), a labelled premium tier, not list."
      },
      "task_copy": {
        "research": {
          "verdict": "The strongest sourced reasoner in the data — 93.9 GPQA (Epoch-run) with 1M context — at half of Fable's price, though still premium per token.",
          "tips": [
            "Start at 'high' effort for analysis and escalate only if the answer is visibly incomplete — effort is the main cost lever.",
            "Give it the full problem and source material in one message; it's built for long single-pass reasoning."
          ]
        }
      },
      "best_for_line": "Anthropic model · 1M ctx · $5/$25",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-opus-5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/about-claude/pricing"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 95,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 94,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 40,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 89,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 53,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 61,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 95,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.22,
          "rank": 46,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": 5,
          "usage_share": 6.53,
          "arena_rank": 11,
          "expert_default": true,
          "families": 3
        },
        "agents": {
          "usage_rank": 6,
          "usage_share": 4.62,
          "arena_rank": 2,
          "expert_default": true,
          "families": 3
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 12,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 1,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 2,
          "usage_share": 7.73,
          "arena_rank": 9,
          "expert_default": null,
          "families": 2
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 9,
          "expert_default": true,
          "families": 2
        },
        "frontend": {
          "usage_rank": 8,
          "usage_share": 2.76,
          "arena_rank": 3,
          "expert_default": true,
          "families": 3
        },
        "exec-summaries": {
          "usage_rank": 9,
          "usage_share": 2.97,
          "arena_rank": 4,
          "expert_default": null,
          "families": 2
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 9,
              "n_models": 26,
              "score": 0.466,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 5 rows)",
              "rank": 26,
              "n_models": 140,
              "score": 0.5637,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 6,
              "n_models": 125,
              "score": 2164.57,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 3 rows)",
              "rank": 4,
              "n_models": 153,
              "score": 0.9178,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 3,
              "n_models": 26,
              "score": 0.7365,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 6,
            "share": 5,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 11,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 5,
              "n_models": 66,
              "score": 11181.8733,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 9,
              "n_models": 56,
              "score": 0.658,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OSWorld 2.0 (Epoch mirror) (best rung: max)",
              "rank": 1,
              "n_models": 10,
              "score": 0.3143,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/osworld-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 6,
            "share": 4.48,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 2,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 62,
            "share": 0.19,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 11,
            "share": 0.55,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 12,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 15,
              "n_models": 295,
              "score": 0.9388,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 17,
              "n_models": 89,
              "score": 0.599,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 13,
              "n_models": 113,
              "score": 0.8561,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 13,
              "n_models": 70,
              "score": 0.7317,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 4,
              "n_models": 83,
              "score": 0.99,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 23,
              "n_models": 274,
              "score": 0.9889,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 12,
              "n_models": 170,
              "score": 0.2914,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 7,
              "n_models": 193,
              "score": 0.9042,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 8,
            "share": 2.49,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 15,
              "n_models": 42,
              "score": 0.24,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 11,
            "share": 1.51,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 1,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 2,
            "share": 5.03,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 9,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 9,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 8,
            "share": 2.76,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 7,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 9,
            "share": 2.97,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena rank matrix, \"Longer Query\" column (proxy — Arena has no dedicated summarization category)",
            "rank": 4,
            "n_models": 5,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        }
      }
    },
    {
      "id": "kimi-k3",
      "name": "Kimi K3",
      "vendor": "Moonshot AI",
      "released": "2026-07-16",
      "context_window": 1000000,
      "price_input": 3,
      "price_output": 15,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": 93.1,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "long-context",
        "vision"
      ],
      "strengths": [
        "Front-end champion — #1 in the Frontend Code Arena (1679 Elo), ahead of Claude Fable 5, winning 6 of 7 frontend categories in blind dev testing",
        "2.8T-param MoE (~16 of 896 experts active per token) — the largest open-weight model to date; 1M-token context with native vision",
        "Cheap for its tier at $3/$15 per M, with ~$0.30/M cache-hit input (90%+ cache-hit rates typical in coding loops)"
      ],
      "weaknesses": [
        "Its headline win is frontend/web-dev coding; SWE-bench Verified is not yet published, so overall coding is a frontend-led estimate (GPQA is now independently published by Epoch: 93.1)",
        "Open weights were slated for ~2026-07-27 and haven't been re-verified as shipped here — treat self-hosting as unconfirmed until you check",
        "Output at $15/M is well above the cheap open tiers (DeepSeek/Qwen) for bulk work"
      ],
      "verdict": "Reach for it for frontend and web-dev coding, where it tops the arena ahead of even Fable 5 at a fraction of the price — but treat its broader coding strength as unproven until SWE-bench lands.",
      "confidence": "medium",
      "sources": [
        "https://www.tomshardware.com/tech-industry/artificial-intelligence/moonshot-releases-2-8-trillion-parameter-kimi-k3",
        "https://openrouter.ai/moonshotai/kimi-k3",
        "https://benchlm.ai/models/kimi-3",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": 96,
      "coding_basis": "Frontend Code Arena #1 (1679 Elo), ahead of Fable 5, winning 6 of 7 frontend categories; general SWE-bench Verified not yet published — score is frontend-led.",
      "coding_confidence": "medium",
      "use_well": [
        "Reach for it on frontend and web-UI builds — that's where it tops the arena; treat broader coding as unproven until SWE-bench lands.",
        "Coding loops hit ~$0.30/M cache-hit input — keep one long session going instead of restarting, and the real cost drops well below sticker.",
        "Planning to self-host? Its open-weight release (slated 2026-07-27) hasn't been re-verified as shipped — confirm before you build around it."
      ],
      "task_copy": {
        "research": {
          "verdict": "A frontier-class reasoner on paper — 93.1 GPQA, independently run by Epoch — at mid-tier prices; its broader benchmark picture is still young.",
          "tips": [
            "Its GPQA figure is Epoch-run at stated effort — treat it as best-case and give hard reasoning a generous token budget.",
            "1M context makes it a strong fit for reasoning across large document sets in one pass."
          ]
        }
      },
      "best_for_line": "Priciest Moonshot AI model · 1M ctx · $3/$15",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "moonshotai/kimi-k3",
        "aws_bedrock": true,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/current/us-east-1/index.json"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 96,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 90,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 45,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 88,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 56,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 64,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": 96,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.04,
          "rank": 21,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": 2,
          "usage_share": 9.08,
          "arena_rank": 6,
          "expert_default": null,
          "families": 2
        },
        "agents": {
          "usage_rank": 3,
          "usage_share": 6.21,
          "arena_rank": 7,
          "expert_default": null,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": 5,
          "usage_share": 3.96,
          "arena_rank": 25,
          "expert_default": null,
          "families": 2
        },
        "research": {
          "usage_rank": 2,
          "usage_share": 7.74,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 4,
          "usage_share": 3.87,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 8,
          "usage_share": 2.4,
          "arena_rank": 12,
          "expert_default": null,
          "families": 2
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 2,
          "usage_share": 7.43,
          "arena_rank": 5,
          "expert_default": null,
          "families": 2
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 5,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 3 rows)",
              "rank": 12,
              "n_models": 140,
              "score": 0.5949,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 24,
              "n_models": 125,
              "score": 1524.47,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 9,
              "n_models": 153,
              "score": 0.8257,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 8,
              "n_models": 26,
              "score": 0.6851,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 3,
            "share": 8.37,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 6,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 31,
              "n_models": 66,
              "score": 5165.035,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 33,
              "n_models": 56,
              "score": 0.478,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 3,
            "share": 6.6,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 7,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 31,
            "share": 0.63,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 5,
            "share": 3.59,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 25,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 21,
              "n_models": 295,
              "score": 0.9312,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 27,
              "n_models": 89,
              "score": 0.506,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 25,
              "n_models": 113,
              "score": 0.7218,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 27,
              "n_models": 70,
              "score": 0.3902,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 11,
              "n_models": 83,
              "score": 0.87,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 35,
              "n_models": 274,
              "score": 0.9722,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 24,
              "n_models": 170,
              "score": 0.234,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 47,
              "n_models": 193,
              "score": 0.6042,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 2,
            "share": 12.65,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 24,
              "n_models": 42,
              "score": 0.19,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 9,
            "share": 1.73,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 9,
            "share": 1.29,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 12,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 2,
            "share": 7.43,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 9,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena rank matrix, \"Longer Query\" column (proxy — Arena has no dedicated summarization category)",
            "rank": 5,
            "n_models": 5,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        }
      }
    },
    {
      "id": "claude-fable-5",
      "name": "Claude Fable 5",
      "vendor": "Anthropic",
      "released": "2026-07",
      "context_window": 1000000,
      "price_input": 10,
      "price_output": 50,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 95.5,
        "gpqa": 91.3,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "agentic",
        "coding",
        "research",
        "long-context"
      ],
      "strengths": [
        "Highest SWE-bench Verified on the board (~95.5% self-announced, ~95% independent)",
        "Anthropic's most capable widely-available model; built for long-horizon autonomous agents",
        "1M-token context at standard pricing",
        "Always-on thinking; high-resolution vision"
      ],
      "weaknesses": [
        "Most expensive model here: $10/$50 per M — exactly 2x Opus 4.8",
        "Reported availability/export restrictions (was briefly pulled offline June 2026)",
        "Requires 30-day data retention; not available under zero-data-retention",
        "Overkill and overpriced for routine work"
      ],
      "verdict": "Use only for the hardest reasoning or overnight autonomous agent runs where its capability edge justifies paying 2x Opus.",
      "confidence": "medium",
      "sources": [
        "https://platform.claude.com/docs/en/about-claude/pricing",
        "https://www.vals.ai/benchmarks/swebench",
        "https://www.vellum.ai/blog/claude-fable-5-and-mythos-5-benchmarks-explained",
        "https://llm-stats.com/benchmarks/swe-bench-verified",
        "https://arena.ai/leaderboard/text",
        "https://www.anthropic.com/news/claude-opus-5",
        "https://arena.ai/leaderboard/agent",
        "https://docs.claude.com/en/docs/claude-code/model-config",
        "https://arena.ai/leaderboard/search",
        "https://arena.ai/leaderboard/document",
        "https://arena.ai/leaderboard/vision",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": 95,
      "coding_basis": "SWE-bench Verified 95.5% (LMArena Code Elo #1)",
      "coding_confidence": "high",
      "use_well": [
        "Save it for the hardest overnight agent runs and deepest reasoning — for everyday coding, Opus 4.8 or Sonnet 5 do the job at a fraction of the price.",
        "Thinking is always on — give it the complete task up front in one well-specified message and let it run; drip-feeding steps wastes its strength.",
        "It's built for 1M-token context: paste whole codebases or document sets in rather than pre-summarizing."
      ],
      "best_for_line": "Anthropic model · 1M ctx · $10/$50",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-fable-5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/about-claude/pricing"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 95,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 94,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 32,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 91,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 87,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 46,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 55,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 95,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 91,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Code model-config docs name Claude Fable 5 (with Fable 5.1) for tasks larger than a single sitting, describing both as sustaining long autonomous sessions, checked 2026-09-06.",
              "source_url": "https://docs.claude.com/en/docs/claude-code/model-config",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Claude Fable 5.1 and Claude Fable 5 are the most capable models in Claude Code, suited to tasks larger than a single sitting."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "writing": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "research": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "extraction": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "chat": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "vision": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "frontend": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "exec-summaries": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Opus 5 launch post (Jul 24, 2026) describes Claude Fable 5 as its frontier-intelligence model, with Opus 5 pitched as approaching it at half the price, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-opus-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "comes close to the frontier intelligence of Claude Fable 5 at half the price"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.02,
          "rank": 131,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 3,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 4,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 1,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 5,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 4,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 1,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 1,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 8,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 1,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 5,
              "n_models": 140,
              "score": 0.61,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 10,
              "n_models": 125,
              "score": 2041.3,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 3,
              "n_models": 153,
              "score": 0.9194,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: xhigh)",
              "rank": 5,
              "n_models": 26,
              "score": 0.6991,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 3,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror) (best of 5 rows)",
              "rank": 26,
              "n_models": 66,
              "score": 5680.258,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 13,
              "n_models": 56,
              "score": 0.636,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 4,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 190,
            "share": 0.01,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 1,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 78,
              "n_models": 295,
              "score": 0.8586,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 7,
              "n_models": 89,
              "score": 0.707,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 10,
              "n_models": 113,
              "score": 0.8702,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 5,
              "n_models": 70,
              "score": 0.902,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 8,
              "n_models": 83,
              "score": 0.95,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 11,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 13,
              "n_models": 170,
              "score": 0.2857,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 10,
              "n_models": 193,
              "score": 0.8917,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Search Arena (34 models, 1,110,523 votes) — grounded web search, closest proxy; Arena has no dedicated research/analysis category",
            "rank": 5,
            "n_models": 34,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/search"
          }
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 7,
              "n_models": 42,
              "score": 0.3,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 4,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 1,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 1,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 13,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena rank matrix, \"Longer Query\" column (proxy — Arena has no dedicated summarization category)",
            "rank": 1,
            "n_models": 5,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        }
      }
    },
    {
      "id": "claude-opus-4-8",
      "name": "Claude Opus 4.8",
      "vendor": "Anthropic",
      "released": "2026-05",
      "context_window": 1000000,
      "price_input": 5,
      "price_output": 25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 88.6,
        "gpqa": 93.6,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning",
        "writing",
        "long-context"
      ],
      "strengths": [
        "Top generally-available coding model (88.6% SWE-bench Verified)",
        "Flagship Opus intelligence at half Fable's price ($5/$25)",
        "1M context at flat rate; optional Fast mode",
        "GPQA Diamond 93.6% (third-party reported)"
      ],
      "weaknesses": [
        "Premium-priced for high-volume/bulk workloads",
        "Slower and pricier than Sonnet 5 for most production agents",
        "Beaten on paper by the restricted Fable/Mythos tier"
      ],
      "verdict": "The default pick for demanding coding, agentic, and reasoning work when you want top-tier quality without Fable's premium.",
      "confidence": "high",
      "sources": [
        "https://platform.claude.com/docs/en/about-claude/pricing",
        "https://llm-stats.com/benchmarks/swe-bench-verified",
        "https://www.vellum.ai/blog/claude-opus-4-8-benchmarks-explained",
        "https://arena.ai/leaderboard/text",
        "https://www.anthropic.com/news/claude-sonnet-5",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/document"
      ],
      "coding_score": 89,
      "coding_basis": "SWE-bench Verified 88.6% (AA Coding Index 74.3)",
      "coding_confidence": "high",
      "use_well": [
        "The sensible Claude default for hard coding and agent work — step up to Fable only when Opus demonstrably falls short.",
        "Leave adaptive thinking on for multi-step problems — the model decides when and how deeply to think; there's no budget to tune.",
        "Re-sending the same big context (docs, a codebase)? Prompt caching serves it at ~10% of input price — put the stable part first in the prompt."
      ],
      "price_checked": {
        "url": "https://platform.claude.com/docs/en/about-claude/pricing",
        "date": "2026-07-29",
        "input": 5,
        "output": 25,
        "aggregator_note": "OpenRouter shows ~$50 — Fast mode ($10/$50), not list."
      },
      "task_copy": {
        "writing": {
          "verdict": "The strongest general model the data tags for prose — long-form drafting, tone control, and editing at flagship quality, at flagship output prices.",
          "tips": [
            "Draft long-form in one pass: give it the full brief and structure up front rather than feeding it paragraph by paragraph.",
            "For rewrites, paste the original plus one sentence of direction — it holds a voice well across long documents."
          ]
        }
      },
      "best_for_line": "Anthropic model · 1M ctx · $5/$25",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-opus-4.8",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/about-claude/pricing"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 89,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 90,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 40,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 88,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 52,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 61,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 89,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Sonnet 5 launch post recommends Claude Opus 4.8 specifically for cybersecurity work needing reduced guardrails, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-sonnet-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "we recommend Claude Opus 4.8 for cybersecurity work that requires reduced guardrails."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Sonnet 5 launch post recommends Claude Opus 4.8 specifically for cybersecurity work needing reduced guardrails, checked 2026-09-06 -- general corroborating evidence of a recognized strength, not task-specific.",
              "source_url": "https://www.anthropic.com/news/claude-sonnet-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "we recommend Claude Opus 4.8 for cybersecurity work that requires reduced guardrails."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "extraction": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Sonnet 5 launch post recommends Claude Opus 4.8 specifically for cybersecurity work needing reduced guardrails, checked 2026-09-06 -- general corroborating evidence of a recognized strength, not task-specific.",
              "source_url": "https://www.anthropic.com/news/claude-sonnet-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "we recommend Claude Opus 4.8 for cybersecurity work that requires reduced guardrails."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.11,
          "rank": 67,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 8,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 6,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 19,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 12,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 12,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 18,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 12,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 40,
              "n_models": 140,
              "score": 0.5347,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror) (best of 2 rows)",
              "rank": 23,
              "n_models": 125,
              "score": 1563.83,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 3 rows)",
              "rank": 8,
              "n_models": 153,
              "score": 0.8289,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 14,
              "n_models": 26,
              "score": 0.5897,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 15,
            "share": 0.06,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 8,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror) (best of 2 rows)",
              "rank": 25,
              "n_models": 66,
              "score": 5787.428,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 31,
              "n_models": 56,
              "score": 0.489,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OSWorld 2.0 (Epoch mirror) (best rung: max)",
              "rank": 3,
              "n_models": 10,
              "score": 0.206,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/osworld-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 6,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 89,
            "share": 0.08,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 19,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 31,
              "n_models": 295,
              "score": 0.9104,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 23,
              "n_models": 89,
              "score": 0.53,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepResearch Bench (Epoch mirror) (best of 3 rows)",
              "rank": 10,
              "n_models": 36,
              "score": 0.502,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepresearchbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 18,
              "n_models": 113,
              "score": 0.8,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 20,
              "n_models": 70,
              "score": 0.561,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 19,
              "n_models": 83,
              "score": 0.69,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 28,
              "n_models": 274,
              "score": 0.9833,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 26,
              "n_models": 170,
              "score": 0.2086,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 13,
            "share": 0.47,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Search Arena (34 models, 1,110,523 votes) — grounded web search, closest proxy; Arena has no dedicated research/analysis category",
            "rank": 12,
            "n_models": 34,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/search"
          }
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 14,
              "n_models": 42,
              "score": 0.24,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 12,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 18,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 12,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 29,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "claude-sonnet-5",
      "name": "Claude Sonnet 5",
      "vendor": "Anthropic",
      "released": "2026-06-30",
      "context_window": 1000000,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 85.2,
        "gpqa": 90.5,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "cheap-bulk",
        "speed",
        "long-context"
      ],
      "strengths": [
        "~85% SWE-bench Verified at a fraction of Opus's price — best coding value right now",
        "1M context; default model on Free/Pro plans",
        "Intro pricing $2/$10 per M through Aug 31 2026",
        "Built to run agents cheaply at scale"
      ],
      "weaknesses": [
        "Intro price rises to standard $3/$15 on Sept 1 2026",
        "Below Opus/Fable on the hardest coding and reasoning tasks",
        "Some third-party benchmark figures for it are inconsistent across sources"
      ],
      "verdict": "The cost-efficient workhorse: most of Opus's coding ability at ~40% of the cost — the daily driver for personal projects.",
      "confidence": "high",
      "sources": [
        "https://platform.claude.com/docs/en/about-claude/pricing",
        "https://llm-stats.com/benchmarks/swe-bench-verified",
        "https://techcrunch.com/2026/06/30/anthropic-launches-claude-sonnet-5-as-a-cheaper-way-to-run-agents/",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/document",
        "https://arena.ai/leaderboard/vision",
        "https://arena.ai/leaderboard/code/webdev",
        "https://roboflow.com/blog/claude-sonnet-5-for-vision"
      ],
      "coding_score": 85,
      "coding_basis": "SWE-bench Verified 85.2% (AA Coding Index 71.5)",
      "coding_confidence": "high",
      "use_well": [
        "The daily driver: ~85 coding at $2/$10 — use it for routine coding and agents, and escalate to Opus only for the genuinely hard problems.",
        "Intro pricing ends Aug 31, 2026 (then $3/$15) — budget for the increase if you're building on it.",
        "Built to run agents cheaply at scale — a good default for CI bots and background automations."
      ],
      "best_for_line": "Anthropic model · 1M ctx · $2/$10",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-sonnet-5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/about-claude/pricing"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 85,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 82,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 49,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 91,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 86,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 58,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 65,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 90,
          "basis": [
            "coding_score",
            "best_for:speed"
          ]
        },
        "exec-summaries": {
          "score": 91,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "vision": {
          "claims": [
            {
              "sentence": "Roboflow's independent vision evaluation (67 prompts, Jul 1, 2026) found Claude Sonnet 5 scoring 67% on document understanding, the weakest result of the four models it tested, checked 2026-10-30.",
              "source_url": "https://roboflow.com/blog/claude-sonnet-5-for-vision",
              "tier": "reported",
              "date": "2026-10-30",
              "quote": "Document understanding drops to 67%, the weakest of the four.",
              "polarity": "negative"
            },
            {
              "sentence": "The same Roboflow evaluation (Jul 1, 2026) found Claude Sonnet 5 scoring 20% (2 of 10) on object counting, the lowest score among the models tested, checked 2026-10-30.",
              "source_url": "https://roboflow.com/blog/claude-sonnet-5-for-vision",
              "tier": "reported",
              "date": "2026-10-30",
              "quote": "And object counting falls to 20% (2 of 10), the lowest score in the group.",
              "polarity": "negative"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-30"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.29,
          "rank": 40,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 24,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 8,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": 7,
          "usage_share": 3.72,
          "arena_rank": 53,
          "expert_default": true,
          "families": 3
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 14,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 4,
          "usage_share": 4.06,
          "arena_rank": 48,
          "expert_default": true,
          "families": 3
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 29,
          "expert_default": true,
          "families": 2
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 23,
              "n_models": 26,
              "score": 0.341,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 36,
              "n_models": 140,
              "score": 0.5428,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 27,
              "n_models": 125,
              "score": 1463.12,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 25,
              "n_models": 153,
              "score": 0.6878,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 17,
              "n_models": 26,
              "score": 0.5385,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 14,
            "share": 0.06,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 24,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 21,
              "n_models": 66,
              "score": 6377.7017,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 23,
              "n_models": 56,
              "score": 0.545,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 8,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 58,
            "share": 0.21,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 7,
            "share": 3.36,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 53,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 2 rows)",
              "rank": 40,
              "n_models": 295,
              "score": 0.9053,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified (best of 2 rows)",
              "rank": 60,
              "n_models": 89,
              "score": 0.337,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 37,
              "n_models": 113,
              "score": 0.6561,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 36,
              "n_models": 70,
              "score": 0.2927,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 15,
              "n_models": 83,
              "score": 0.77,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 2 rows)",
              "rank": 47,
              "n_models": 274,
              "score": 0.9472,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 40,
              "n_models": 170,
              "score": 0.1686,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 10,
            "share": 2.4,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": {
            "rank": 10,
            "share": 1.65,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 14,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 6,
            "share": 2.07,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 48,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 29,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 38,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gpt-5-5",
      "name": "GPT-5.5",
      "vendor": "OpenAI",
      "released": "2026-04-23",
      "context_window": 1000000,
      "price_input": 5,
      "price_output": 30,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 81.9,
        "gpqa": 93.6,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "reasoning",
        "agentic",
        "research"
      ],
      "strengths": [
        "Previous OpenAI flagship; still a strong all-rounder",
        "SWE-bench Verified 81.9% (reported), GPQA Diamond 93.6%",
        "LMArena text Elo 1475; 1M-token API context"
      ],
      "weaknesses": [
        "Superseded by GPT-5.6 at the same price — usually no reason to pick over Sol/Terra now",
        "Sources conflict on its coding metric (Verified 81.9% vs SWE-bench Pro 58.6%)",
        "Weaker multimodal performance"
      ],
      "verdict": "Only choose over GPT-5.6 if you've already validated a pipeline on it; otherwise 5.6 dominates at equal cost.",
      "confidence": "medium",
      "sources": [
        "https://benchlm.ai/models/gpt-5-5",
        "https://openai.com/index/introducing-gpt-5-5/",
        "https://felloai.com/the-ultimate-chatgpt-model-comparison/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/search",
        "https://arena.ai/leaderboard/document"
      ],
      "coding_score": 82,
      "coding_basis": "SWE-bench Verified 81.9% + AA Coding Index 74.9",
      "coding_confidence": "high",
      "use_well": [
        "Superseded by GPT-5.6 at the same price — starting fresh on OpenAI, pick Terra (value) or Sol (peak) instead.",
        "Only stay on it if a validated pipeline is already pinned to it."
      ],
      "price_checked": {
        "url": "https://developers.openai.com/api/docs/pricing",
        "date": "2026-07-29",
        "input": 5,
        "output": 30,
        "aggregator_note": "OpenRouter shows ~$180 — matches no tier OpenAI publishes."
      },
      "best_for_line": "OpenAI model · 1M ctx · $5/$30",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "openai/gpt-5.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://developers.openai.com/api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 82,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 85,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 38,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 88,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 51,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 59,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 82,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.02,
          "rank": 118,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 28,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 9,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 28,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 7,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 19,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 11,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 27,
              "n_models": 140,
              "score": 0.5613,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror) (best of 3 rows)",
              "rank": 12,
              "n_models": 125,
              "score": 1942.97,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 3 rows)",
              "rank": 6,
              "n_models": 153,
              "score": 0.8491,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: xhigh)",
              "rank": 11,
              "n_models": 26,
              "score": 0.6704,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 28,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Terminal-Bench 2.0 (Epoch mirror) (best rung: NexAU-AHE)",
              "rank": 1,
              "n_models": 58,
              "score": 0.8472,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/terminalbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 17,
              "n_models": 66,
              "score": 7523.844,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 21,
              "n_models": 56,
              "score": 0.551,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OSWorld 2.0 (Epoch mirror) (best rung: xhigh)",
              "rank": 5,
              "n_models": 10,
              "score": 0.13,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/osworld-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 9,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 115,
            "share": 0.04,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 28,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 38,
              "n_models": 295,
              "score": 0.9066,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 14,
              "n_models": 89,
              "score": 0.63,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepResearch Bench (Epoch mirror) (best of 3 rows)",
              "rank": 4,
              "n_models": 36,
              "score": 0.54,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepresearchbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2 (best of 2 rows)",
              "rank": 14,
              "n_models": 113,
              "score": 0.8526,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2 (best of 2 rows)",
              "rank": 14,
              "n_models": 70,
              "score": 0.725,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: xhigh)",
              "rank": 31,
              "n_models": 83,
              "score": 0.5,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 89,
              "n_models": 274,
              "score": 0.8444,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 16,
              "n_models": 170,
              "score": 0.2714,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 12,
              "n_models": 193,
              "score": 0.85,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 12,
              "n_models": 42,
              "score": 0.26,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 7,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 19,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 11,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 45,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gemini-3-1-pro",
      "name": "Gemini 3.1 Pro (Preview)",
      "vendor": "Google",
      "released": "2026",
      "context_window": 1000000,
      "price_input": 2,
      "price_output": 12,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 80.6,
        "gpqa": 94.3,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "coding",
        "agentic",
        "research",
        "vision",
        "long-context"
      ],
      "strengths": [
        "Current Gemini flagship; top-tier reasoning (GPQA 94.3), tied at the top of the AA Intelligence Index",
        "Strong agentic coding (SWE-bench Verified 80.6)",
        "Cheap for a frontier model ($2/$12 up to 200k tokens); 1M context, strong multimodal"
      ],
      "weaknesses": [
        "Price doubles to $4/$18 above 200k-token prompts",
        "Still a Preview SKU (may change)",
        "Trails top Claude models on coding"
      ],
      "verdict": "The best-value frontier coder outside Anthropic — reach for it for hard reasoning, big multimodal context, or the Google ecosystem.",
      "confidence": "high",
      "sources": [
        "https://ai.google.dev/gemini-api/docs/pricing",
        "https://llm-stats.com/benchmarks/swe-bench-verified",
        "https://smartchunks.com/gemini-3-1-pro-benchmarks-gpqa-hle-lmsys-frontiermath/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/agent",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/search",
        "https://arena.ai/leaderboard/document",
        "https://arena.ai/leaderboard/vision"
      ],
      "coding_score": 81,
      "coding_basis": "SWE-bench Verified 80.6% + AA Coding Index 68.8",
      "coding_confidence": "high",
      "use_well": [
        "Strongest budget pick for hard reasoning and big multimodal jobs (documents + images together).",
        "Watch the 200K-token threshold — price doubles to $4/$18 above it, so chunk very long inputs when you can.",
        "Still a Preview SKU — pin the version and re-test before production bets."
      ],
      "best_for_line": "Priciest Google model · 1M ctx · $2/$12",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "google/gemini-3.1-pro-preview",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://ai.google.dev/gemini-api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 81,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 80,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 48,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 89,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 59,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 66,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": 81,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 94,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.05,
          "rank": 88,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "preview",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 25,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 43,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 10,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 24,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 15,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 24,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 16,
              "n_models": 140,
              "score": 0.5868,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 48,
              "n_models": 125,
              "score": 1160.6,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 22,
              "n_models": 153,
              "score": 0.7207,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: high)",
              "rank": 26,
              "n_models": 26,
              "score": 0.1173,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 25,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Terminal-Bench 2.0 (Epoch mirror) (best rung: TongAgents)",
              "rank": 3,
              "n_models": 58,
              "score": 0.8022,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/terminalbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 53,
              "n_models": 66,
              "score": 911.214,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "METR time horizons (Epoch mirror)",
              "rank": 3,
              "n_models": 46,
              "score": 384.1474,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/metr-time-horizons-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 47,
              "n_models": 56,
              "score": 0.351,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 43,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 52,
            "share": 0.26,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 12,
            "share": 0.3,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 10,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 2 rows)",
              "rank": 8,
              "n_models": 295,
              "score": 0.9444,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 4,
              "n_models": 89,
              "score": 0.735,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 3,
              "n_models": 50,
              "score": 0.4644,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepResearch Bench (Epoch mirror) (best of 2 rows)",
              "rank": 19,
              "n_models": 36,
              "score": 0.478,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepresearchbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 41,
              "n_models": 113,
              "score": 0.5965,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 41,
              "n_models": 70,
              "score": 0.2683,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: high)",
              "rank": 46,
              "n_models": 83,
              "score": 0.26,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 2 rows)",
              "rank": 40,
              "n_models": 274,
              "score": 0.956,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 37,
              "n_models": 170,
              "score": 0.1771,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 21,
              "n_models": 193,
              "score": 0.7708,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 27,
              "n_models": 42,
              "score": 0.17,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 8,
            "share": 2.79,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 24,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 8,
            "share": 1.45,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 15,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 24,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 59,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "deepseek-v4-pro",
      "name": "DeepSeek V4-Pro",
      "vendor": "DeepSeek",
      "released": "2026-04-24",
      "context_window": 1000000,
      "price_input": 1.32,
      "price_output": 3.96,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 80.6,
        "gpqa": 90.1,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "reasoning",
        "long-context",
        "agentic"
      ],
      "strengths": [
        "Near-Opus coding quality (80.6% SWE-bench Verified) at a fraction of the price",
        "Open weights (MIT); self-hostable",
        "1M-token context; cache-hit input extraordinarily cheap (~$0.0036/M)"
      ],
      "weaknesses": [
        "Top SWE-bench score is 'Think Max' mode (slower/pricier in tokens)",
        "Benchmark figures from secondary aggregators, not a primary vendor card",
        "Trails best closed models by ~8 pts on coding"
      ],
      "verdict": "The best-value open-weight coder — frontier-ish quality without frontier prices, and self-hostable to avoid lock-in.",
      "confidence": "high",
      "sources": [
        "https://api-docs.deepseek.com/quick_start/pricing/",
        "https://llm-stats.com/benchmarks/swe-bench-verified",
        "https://benchlm.ai/models/deepseek-v4-pro-max",
        "https://raw.githubusercontent.com/BerriAI/litellm/main/model_prices_and_context_window.json",
        "https://api-docs.deepseek.com/quick_start/pricing",
        "https://www.buildfastwithai.com/blogs/deepseek-v4-pro-review-2026",
        "https://artificialanalysis.ai/models/deepseek-v4-pro",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 81,
      "coding_basis": "SWE-bench Verified 80.6% (AA Coding Index lower, 59.4)",
      "coding_confidence": "medium",
      "use_well": [
        "Near-frontier coding at a fraction of frontier price — try it first for cost-sensitive coding before paying frontier rates.",
        "Cache-hit input is nearly free (~$0.0036/M) — agent loops that resend the same prefix cost almost nothing.",
        "Its top benchmark scores use 'Think Max' mode — slower and token-hungry; save that mode for hard problems, not bulk."
      ],
      "task_copy": {
        "research": {
          "verdict": "The value pick for strategy and analysis: 90.1 GPQA — within a few points of the frontier — at open-weight prices, and self-hostable to avoid lock-in.",
          "tips": [
            "Use its 'Think' mode for hard analysis and planning; standard mode handles routine synthesis at a fraction of the tokens.",
            "Cache-hit input is nearly free (~$0.0036/M) — long research sessions that keep building on the same context cost almost nothing."
          ]
        }
      },
      "best_for_line": "Priciest DeepSeek model · 1M ctx · $1.32/$3.96",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://api-docs.deepseek.com/quick_start/pricing/",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 81,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 80,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 58,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 90,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 86,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 65,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 70,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 81,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 90,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "llm-stats.com's SWE-Bench Verified leaderboard (checked 2026-10-16) lists DeepSeek's max-reasoning-effort configuration of V4-Pro at a 0.806 score, rank 9 of the 117 models it tracks.",
              "source_url": "https://llm-stats.com/benchmarks/swe-bench-verified",
              "tier": "reported",
              "date": "2026-10-16",
              "quote": "DeepSeek-V4-Pro-Max by DeepSeek is the top-ranked open-source model on SWE-Bench Verified, with a score of 0.806 (rank #9)."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-16"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.32,
          "rank": 38,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 56,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 32,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 45,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 52,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SWE-bench Verified (Epoch mirror)",
              "rank": 6,
              "n_models": 33,
              "score": 0.7764,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/swe-bench-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 4 rows)",
              "rank": 55,
              "n_models": 140,
              "score": 0.5,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror) (best of 2 rows)",
              "rank": 56,
              "n_models": 125,
              "score": 1006.08,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 61,
              "n_models": 153,
              "score": 0.489,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 56,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 46,
              "n_models": 66,
              "score": 3284.52,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 32,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 36,
            "share": 0.54,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 45,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 36,
              "n_models": 295,
              "score": 0.9091,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 36,
              "n_models": 89,
              "score": 0.4699,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 57,
              "n_models": 113,
              "score": 0.4526,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 64,
              "n_models": 70,
              "score": 0.0244,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 60,
              "n_models": 83,
              "score": 0.16,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 36,
              "n_models": 274,
              "score": 0.9667,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 51,
              "n_models": 170,
              "score": 0.1286,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 52,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 56,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "kimi-k2-6",
      "name": "Kimi K2.6",
      "vendor": "Moonshot AI",
      "released": "2026-04-20",
      "context_window": 262144,
      "price_input": 0.95,
      "price_output": 4,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 76.7,
        "gpqa": 90.8,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "long-context"
      ],
      "strengths": [
        "Elite agentic coding — 80.2% SWE-bench Verified; ties GPT-5.5 on SWE-bench Pro",
        "1T-param MoE (32B active), open weights",
        "Built for long-horizon multi-step agent tasks"
      ],
      "weaknesses": [
        "Pricier output than the cheapest DeepSeek/Qwen tiers",
        "Third-party host prices vary widely ($1.15-$2.15 blended)",
        "262K context, below the 1M-token models"
      ],
      "verdict": "Reach for it when building coding agents that run long multi-step tasks and you want open weights that rival top closed models.",
      "confidence": "high",
      "sources": [
        "https://platform.kimi.ai/docs/pricing/chat-k26",
        "https://aitoolsrecap.com/Blog/moonshot-ai-kimi-k2-6-release-coding-agent-benchmarks-2026",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://epoch.ai/benchmarks/swe-bench-verified",
        "https://www.kimi.ai/blog/kimi-k2-6",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 77,
      "coding_basis": "SWE-bench Verified 76.7% (Epoch AI independent re-run, 2026-07-25) + LMArena Code Elo #17 (1513). Replaces an 80.2% figure that traced to llm-stats, an aggregator; the independent re-run outranks it.",
      "coding_confidence": "high",
      "use_well": [
        "Best when you want an open-weight, long-horizon coding agent — 80% SWE-bench at open-model prices.",
        "262K context sits below the 1M crowd — for very large codebases pick a 1M-context model instead."
      ],
      "price_checked": {
        "url": "https://platform.kimi.ai/docs/pricing/chat-k26",
        "date": "2026-07-29",
        "input": 0.95,
        "output": 4,
        "aggregator_note": "OpenRouter shows ~$2.72 — provider pass-through. Vendor input is cache-miss $0.95 / cache-hit $0.16."
      },
      "best_for_line": "Cheapest Moonshot AI model · 262K ctx · $0.95/$4",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "moonshotai/kimi-k2.6",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.kimi.ai/docs/pricing/chat-k26"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 77,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 60,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 58,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 91,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 69,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 66,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 71,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 77,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 91,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Moonshot AI's own Kimi K2.6 tech blog (checked 2026-09-07) states the model shows strong improvements in long-horizon coding tasks with generalization across languages including Rust, Go, and Python.",
              "source_url": "https://www.kimi.ai/blog/kimi-k2-6",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Kimi K2.6 shows strong improvements in long-horizon coding tasks, with reliable generalization across programming languages (e.g., Rust, Go, and Python)"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Moonshot AI's Kimi K2.6 blog (checked 2026-09-07) reports measurable gains in real-world reliability including API interpretation and stability during extended tool-using tasks.",
              "source_url": "https://www.kimi.ai/blog/kimi-k2-6",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Kimi K2.6 delivers measurable improvements in real-world reliability: more precise API interpretation, stabler long-running performance, and enhanced safety awareness during extended research tasks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "Moonshot AI's Kimi K2.6 blog (checked 2026-09-07) documents evaluations run with a context length of 262,144 tokens, relevant to processing large inputs in one pass.",
              "source_url": "https://www.kimi.ai/blog/kimi-k2-6",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Unless otherwise specified, all Kimi K2.6 experiments were conducted with temperature = 1.0, top-p = 1.0, and a context length of 262,144 tokens."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.07,
          "rank": 80,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 36,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 36,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 62,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 22,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 50,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 33,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SWE-bench Verified (Epoch mirror)",
              "rank": 10,
              "n_models": 33,
              "score": 0.7665,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/swe-bench-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 41,
              "n_models": 140,
              "score": 0.5347,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 51,
              "n_models": 125,
              "score": 1092.67,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 46,
              "n_models": 153,
              "score": 0.5586,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 36,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 22,
              "n_models": 66,
              "score": 6204.57,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OSWorld 2.0 (Epoch mirror) (best rung: enabled)",
              "rank": 9,
              "n_models": 10,
              "score": 0.046,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/osworld-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 36,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 75,
            "share": 0.13,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 62,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 37,
              "n_models": 295,
              "score": 0.9078,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 55,
              "n_models": 89,
              "score": 0.349,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 47,
              "n_models": 113,
              "score": 0.5719,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 44,
              "n_models": 70,
              "score": 0.2564,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 58,
              "n_models": 83,
              "score": 0.16,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 37,
              "n_models": 274,
              "score": 0.9611,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 64,
              "n_models": 170,
              "score": 0.08,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 35,
              "n_models": 42,
              "score": 0.12,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 22,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 50,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 33,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 46,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "llama-4-maverick",
      "name": "Llama 4 Maverick",
      "vendor": "Meta",
      "released": "2026-Q1",
      "context_window": 1048576,
      "price_input": 0.22,
      "price_output": 0.88,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 74.2,
        "gpqa": 65.8,
        "aime": null,
        "mmlu_pro": 77.1
      },
      "best_for": [
        "cheap-bulk",
        "long-context",
        "coding"
      ],
      "strengths": [
        "Very cheap for its quality via hosts like Fireworks/Together/Groq",
        "1M-token context; open weights, widely available",
        "MMLU-Pro 77.1, SWE-bench 74.2 (secondary-sourced)"
      ],
      "weaknesses": [
        "No single official Meta list price — cost varies 20%+ by host",
        "Trails top open coders (DeepSeek/Kimi/Qwen) on SWE-bench",
        "Benchmarks from secondary aggregators"
      ],
      "verdict": "A cheap open-weight generalist with huge context — good for bulk work and long documents on a widely-hosted model.",
      "confidence": "medium",
      "sources": [
        "https://tokenmix.ai/blog/llama-4-maverick-review",
        "https://benchlm.ai/models/llama-4-maverick",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 74,
      "coding_basis": "SWE-bench Verified 74.2% (secondary source, uncorroborated)",
      "coding_confidence": "medium",
      "use_well": [
        "A cheap open generalist for bulk work and long documents — not the top open coder, so don't pick it for hard coding.",
        "Prices vary 20%+ by host (Fireworks/Together/Groq) — compare hosts before committing volume."
      ],
      "best_for_line": "Meta model · 1M ctx · $0.22/$0.88",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta-llama/llama-4-maverick",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 74,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 76,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 65,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 66,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 70,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 65,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 65,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 74,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 66,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "TokenMix's review (checked 2026-09-07) reports Llama 4 Maverick scoring 91.5% on HumanEval, near the 92.0% it cites for GPT-5.4, and 89.5% on MBPP.",
              "source_url": "https://tokenmix.ai/blog/llama-4-maverick-review",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Yes. Maverick scores 91.5% on HumanEval (near GPT-5.4's 92.0%) and 89.5% on MBPP."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "TokenMix's review (checked 2026-09-07) estimates a 10,000-document/day analysis workload costs $390/month on Maverick via Groq versus the $5,250 it cites for GPT-5.4, a factor relevant to bulk processing.",
              "source_url": "https://tokenmix.ai/blog/llama-4-maverick-review",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Document analysis 10K docs/day: Maverick (Groq) $390 vs GPT-5.4 $5,250"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.01,
          "rank": 171,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 120,
              "n_models": 140,
              "score": 0.3171,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 135,
              "n_models": 153,
              "score": 0.2447,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 166,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Lech Mazur Writing (Epoch mirror)",
              "rank": 46,
              "n_models": 49,
              "score": 6.2,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/lech-mazur-writing-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 42,
              "n_models": 50,
              "score": 0.0568,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 115,
              "n_models": 170,
              "score": 0,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "o4-mini",
      "name": "o4-mini",
      "vendor": "OpenAI",
      "released": null,
      "context_window": null,
      "price_input": 1.1,
      "price_output": 4.4,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": 68.1,
        "gpqa": 81.4,
        "aime": 92.7,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "cheap-bulk",
        "vision",
        "coding"
      ],
      "strengths": [
        "Cheap reasoning model: AIME 2025 92.7% (beats o3), SWE-bench Verified 68.1%",
        "GPQA 81.4%; reasons over images (multimodal)",
        "~10x cheaper inference than o3 for near-o3 results"
      ],
      "weaknesses": [
        "Retired from the ChatGPT UI (Feb 2026); API-only",
        "Older o-series generation, overshadowed by the GPT-5.x reasoning line",
        "GPQA figure may not be the Diamond subset"
      ],
      "verdict": "A cost-efficient math/reasoning pick via API, but the GPT-5.6 line is the modern default.",
      "confidence": "medium",
      "sources": [
        "https://www.datacamp.com/blog/o4-mini",
        "https://llm-stats.com/models/o4-mini",
        "https://openrouter.ai/openai/o4-mini",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 68,
      "coding_basis": "SWE-bench Verified 68.1% (older gen, uncorroborated)",
      "coding_confidence": "medium",
      "use_well": [
        "A cheap math/reasoning pick via API — strong competition-math scores at a fraction of o3-class cost.",
        "API-only now (retired from the ChatGPT UI) and an older generation — the GPT-5.6 line is the modern default."
      ],
      "price_checked": {
        "url": "https://developers.openai.com/api/docs/pricing",
        "date": "2026-07-29",
        "input": 1.1,
        "output": 4.4,
        "aggregator_note": "OpenRouter shows ~$8 — that's OpenAI's Priority tier ($2/$8), not standard."
      },
      "best_for_line": "OpenAI model · $1.1/$4.4",
      "availability": {
        "direct_api": true,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://developers.openai.com/api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 68,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 68,
          "basis": [
            "coding_score"
          ]
        },
        "bulk": {
          "score": 54,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 81,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 81,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "extraction": {
          "score": 60,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 64,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": 68,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 81,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter's model page (checked 2026-09-07) states o4-mini demonstrates competitive reasoning and coding performance on benchmarks including AIME and SWE-bench.",
              "source_url": "https://openrouter.ai/openai/o4-mini",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It supports tool use and demonstrates competitive reasoning and coding performance across benchmarks like AIME (99.5% with Python) and SWE-bench"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "vision": {
          "claims": [
            {
              "sentence": "OpenRouter's model page (checked 2026-09-07) states that despite its smaller size, o4-mini exhibits high accuracy on visual problem-solving benchmarks including MathVista and MMMU.",
              "source_url": "https://openrouter.ai/openai/o4-mini",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Despite its smaller size, o4-mini exhibits high accuracy in STEM tasks, visual problem solving (e.g., MathVista, MMMU), and code editing."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "OpenRouter's model page (checked 2026-09-07) describes o4-mini as especially well-suited for high-throughput scenarios where latency or cost is critical.",
              "source_url": "https://openrouter.ai/openai/o4-mini",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It is especially well-suited for high-throughput scenarios where latency or cost is critical."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 310,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 165,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 179,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 158,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 75,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 69,
              "n_models": 125,
              "score": 826.17,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 52,
              "n_models": 153,
              "score": 0.5256,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 165,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "METR time horizons (Epoch mirror)",
              "rank": 21,
              "n_models": 46,
              "score": 76.509,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/metr-time-horizons-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 270,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Lech Mazur Writing (Epoch mirror)",
              "rank": 33,
              "n_models": 49,
              "score": 7.5,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/lech-mazur-writing-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 179,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 115,
              "n_models": 295,
              "score": 0.7961,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified (best of 2 rows)",
              "rank": 78,
              "n_models": 89,
              "score": 0.196,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror) (best of 2 rows)",
              "rank": 26,
              "n_models": 50,
              "score": 0.1808,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2 (best of 3 rows)",
              "rank": 64,
              "n_models": 113,
              "score": 0.3614,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 62,
              "n_models": 70,
              "score": 0.0488,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 102,
              "n_models": 274,
              "score": 0.8167,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 101,
              "n_models": 170,
              "score": 0.006,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 103,
              "n_models": 193,
              "score": 0.0611,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 158,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 75,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDPval (Epoch mirror)",
              "rank": 8,
              "n_models": 11,
              "score": 0.253,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdpval-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "claude-haiku-4-5",
      "name": "Claude Haiku 4.5",
      "vendor": "Anthropic",
      "released": "2025-10-01",
      "context_window": 200000,
      "price_input": 1,
      "price_output": 5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "speed",
        "coding"
      ],
      "strengths": [
        "Cheapest Claude ($1/$5 per M)",
        "Fastest tier; ideal for classification, routing, extraction, high-volume simple tasks",
        "~$37 to process 10,000 support tickets (Anthropic's own example)"
      ],
      "weaknesses": [
        "Smaller 200K context vs 1M on the larger models",
        "Not for hard reasoning or long-horizon agentic work",
        "No published SWE-bench/GPQA figures from reliable sources"
      ],
      "verdict": "Reach for it when volume and speed matter more than peak intelligence — cheap bulk classification, routing, and extraction.",
      "confidence": "medium",
      "sources": [
        "https://platform.claude.com/docs/en/about-claude/pricing",
        "https://www.anthropic.com/news/claude-haiku-4-5",
        "https://openrouter.ai/anthropic/claude-haiku-4.5",
        "https://arena.ai/leaderboard/document",
        "https://docs.claude.com/en/docs/about-claude/models/choosing-a-model",
        "https://arena.ai/leaderboard/text"
      ],
      "coding_score": 66,
      "coding_basis": "SWE-bench Pro 39.5% + LiveCodeBench 51.1% (small/fast tier)",
      "coding_confidence": "medium",
      "use_well": [
        "Use it when volume and speed beat peak intelligence: classification, routing, extraction.",
        "Pair it as the cheap tier under a smarter model — let Haiku triage the easy 90%, escalate the hard 10%.",
        "200K context — split very long documents rather than forcing them in."
      ],
      "best_for_line": "Cheapest Anthropic model · 200K ctx · $1/$5",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-haiku-4.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/about-claude/pricing"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 66,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 51,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 71,
          "basis": [
            "coding_score",
            "best_for:speed"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "chat": {
          "claims": [
            {
              "sentence": "Anthropic's Claude Haiku 4.5 launch post (Oct 15, 2025) names real-time, low-latency tasks such as chat assistants, customer service agents and pair programming as uses for Haiku 4.5, checked 2026-10-16.",
              "source_url": "https://www.anthropic.com/news/claude-haiku-4-5",
              "tier": "lab",
              "date": "2026-10-16",
              "quote": "real-time, low-latency tasks like chat assistants, customer service agents, or pair programming will appreciate Haiku 4.5's combination of high intelligence and remarkable speed"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-16"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "Anthropic's own Claude Haiku 4.5 launch post (Oct 15, 2025) says it gives similar coding performance to Claude Sonnet 4 at one-third the cost and more than twice the speed, checked 2026-09-06.",
              "source_url": "https://www.anthropic.com/news/claude-haiku-4-5",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "similar levels of coding performance but at one-third the cost and more than twice the speed"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.08,
          "rank": 75,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 96,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 120,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 31,
          "expert_default": true,
          "families": 2
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 131,
          "expert_default": true,
          "families": 2
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 82,
              "n_models": 140,
              "score": 0.4225,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 92,
              "n_models": 125,
              "score": 653.48,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 75,
              "n_models": 153,
              "score": 0.454,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 96,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Terminal-Bench 2.0 (Epoch mirror) (best rung: Goose)",
              "rank": 38,
              "n_models": 58,
              "score": 0.3551,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/terminalbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 57,
              "n_models": 66,
              "score": 458.886,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 67,
            "share": 0.14,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 120,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 2 rows)",
              "rank": 157,
              "n_models": 295,
              "score": 0.7121,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified (best of 2 rows)",
              "rank": 82,
              "n_models": 89,
              "score": 0.132,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepResearch Bench (Epoch mirror)",
              "rank": 25,
              "n_models": 36,
              "score": 0.455,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepresearchbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 2 rows)",
              "rank": 136,
              "n_models": 274,
              "score": 0.6667,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 119,
              "n_models": 170,
              "score": 0,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 118,
              "n_models": 193,
              "score": 0.0403,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 31,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 131,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 102,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gpt-5-6-sol",
      "name": "GPT-5.6 Sol",
      "vendor": "OpenAI",
      "released": "2026-07-09",
      "context_window": 1000000,
      "price_input": 4,
      "price_output": 20,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": 94.6,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning",
        "research",
        "long-context"
      ],
      "strengths": [
        "Flagship of the GPT-5.6 line; tops the Artificial Analysis Coding Index",
        "SWE-bench Pro 64.6% and Terminal-Bench 2.1 88.8% (agentic-coding leader)",
        "GPQA Diamond 94.6%; strong long-context recall (MRCR 91.5%)",
        "Marketed as OpenAI's best model for cybersecurity/long-horizon tasks"
      ],
      "weaknesses": [
        "OpenAI did NOT publish SWE-bench Verified, AIME, or MMLU-Pro for it",
        "Chain-of-thought reasoning means higher latency and token usage",
        "Higher output price ($30) than Claude Opus"
      ],
      "verdict": "Pick this if you're in the OpenAI/Codex ecosystem and want their strongest coding-agent and reasoning tier; cost is secondary.",
      "confidence": "medium",
      "sources": [
        "https://developers.openai.com/api/docs/pricing",
        "https://artificialanalysis.ai/articles/gpt-5-6-has-landed",
        "https://benchlm.ai/models/gpt-5-6-sol",
        "https://techcrunch.com/2026/07/09/openai-launches-its-new-family-of-models-with-gpt-5-6/",
        "https://www.citybiz.co/article/892929/openai-cuts-gpt-5-6-sol-api-pricing-20-for-three-months/",
        "https://developers.openai.com/api/docs/models/gpt-5.6-sol",
        "https://raw.githubusercontent.com/BerriAI/litellm/main/model_prices_and_context_window.json",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/search",
        "https://arena.ai/leaderboard/document"
      ],
      "coding_score": 90,
      "coding_basis": "AA Coding Index #1 (77.4) + LMArena Code Elo #2 (1636)",
      "coding_confidence": "medium",
      "use_well": [
        "OpenAI's top tier for the hardest agentic, coding, and security work — pay for it only when Terra falls short.",
        "Its chain-of-thought reasoning means higher latency and token spend — budget for both."
      ],
      "best_for_line": "OpenAI model · 1M ctx · $4/$20",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "openai/gpt-5.6-sol",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://developers.openai.com/api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 90,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 91,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 43,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 95,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 89,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 55,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 63,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 90,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 95,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.16,
          "rank": 55,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": 8,
          "usage_share": 4.08,
          "arena_rank": 14,
          "expert_default": true,
          "families": 3
        },
        "agents": {
          "usage_rank": 9,
          "usage_share": 3.26,
          "arena_rank": 5,
          "expert_default": true,
          "families": 3
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": 8,
          "usage_share": 2.89,
          "arena_rank": 11,
          "expert_default": null,
          "families": 2
        },
        "research": {
          "usage_rank": 10,
          "usage_share": 2.64,
          "arena_rank": 1,
          "expert_default": null,
          "families": 2
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 11,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 17,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 17,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": 7,
          "usage_share": 3.13,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 14,
              "n_models": 26,
              "score": 0.417,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 18,
              "n_models": 140,
              "score": 0.5775,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 5,
              "n_models": 125,
              "score": 2176.88,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 3 rows)",
              "rank": 5,
              "n_models": 153,
              "score": 0.8943,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 4,
              "n_models": 26,
              "score": 0.7267,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 7,
            "share": 4.62,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 14,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 9,
              "n_models": 66,
              "score": 9619.368,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 29,
              "n_models": 56,
              "score": 0.514,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OSWorld 2.0 (Epoch mirror) (best rung: max)",
              "rank": 2,
              "n_models": 10,
              "score": 0.2734,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/osworld-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 8,
            "share": 3.5,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 5,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 53,
            "share": 0.26,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 8,
            "share": 3.01,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 11,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 16,
              "n_models": 295,
              "score": 0.935,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 9,
              "n_models": 89,
              "score": 0.697,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 7,
              "n_models": 113,
              "score": 0.8912,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2 (best of 2 rows)",
              "rank": 8,
              "n_models": 70,
              "score": 0.8293,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 13,
              "n_models": 83,
              "score": 0.83,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 10,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 2,
              "n_models": 170,
              "score": 0.3229,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 5,
              "n_models": 193,
              "score": 0.925,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 7,
            "share": 2.66,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Search Arena (34 models, 1,110,523 votes) — grounded web search, closest proxy; Arena has no dedicated research/analysis category",
            "rank": 1,
            "n_models": 34,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/search"
          }
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 5,
              "n_models": 42,
              "score": 0.307,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 6,
            "share": 3.37,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 11,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 17,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 17,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 16,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 7,
            "share": 3.13,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      }
    },
    {
      "id": "gpt-5-6-terra",
      "name": "GPT-5.6 Terra",
      "vendor": "OpenAI",
      "released": "2026-07-09",
      "context_window": 1000000,
      "price_input": 2,
      "price_output": 12,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": 93.3,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning",
        "writing"
      ],
      "strengths": [
        "Balanced mid-tier: half Sol's price with close performance",
        "Terminal-Bench 2.1 87.4%, Coding Agent Index 77.4 (just under Sol)",
        "Cost-per-task ~$0.55 vs Sol's ~$1.04; 1M context, strong long-context recall"
      ],
      "weaknesses": [
        "OpenAI omitted GPQA, AIME, SWE-bench Verified, MMLU-Pro for this tier",
        "Not as strong as Sol on the hardest agentic/security tasks"
      ],
      "verdict": "The default everyday GPT-5.6 pick: most of Sol's ability at half the price.",
      "confidence": "medium",
      "sources": [
        "https://developers.openai.com/api/docs/pricing",
        "https://www.vellum.ai/blog/gpt-5-6-benchmarks-explained",
        "https://artificialanalysis.ai/articles/gpt-5-6-has-landed",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/document"
      ],
      "coding_score": 87,
      "coding_basis": "AA Coding Index 76.7",
      "coding_confidence": "medium",
      "use_well": [
        "The everyday GPT-5.6 pick — most of Sol's ability at half the price (~$0.55 vs ~$1.04 per task).",
        "Try Terra first and escalate to Sol only when a task provably needs it."
      ],
      "task_copy": {
        "writing": {
          "verdict": "The value writing pick — near-flagship prose at half Sol's price, suited to everyday drafts, emails, and content.",
          "tips": [
            "Everyday drafts and content are its lane; escalate to a flagship only for high-stakes long-form.",
            "Show it two or three samples of your voice — it imitates provided style more reliably than described style."
          ]
        }
      },
      "price_checked": {
        "url": "https://developers.openai.com/api/docs/pricing",
        "date": "2026-07-30",
        "input": 2,
        "output": 12,
        "aggregator_note": "20% price cut announced 2026-07-30 (was $2.50/$15); confirmed live on OpenAI's own pricing page."
      },
      "best_for_line": "OpenAI model · 1M ctx · $2/$12",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "openai/gpt-5.6-terra",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://developers.openai.com/api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 87,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 89,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 48,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 88,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 58,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 65,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 87,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.13,
          "rank": 63,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 31,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 25,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 68,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 5,
          "usage_share": 3.92,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 10,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 43,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 32,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 18,
              "n_models": 26,
              "score": 0.413,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 32,
              "n_models": 140,
              "score": 0.5498,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 11,
              "n_models": 125,
              "score": 1951.42,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 11,
              "n_models": 153,
              "score": 0.7827,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 6,
              "n_models": 26,
              "score": 0.6962,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 31,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 18,
              "n_models": 66,
              "score": 7343.206,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 17,
              "n_models": 56,
              "score": 0.582,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 11,
            "share": 0.23,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 25,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 80,
            "share": 0.1,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 68,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 18,
              "n_models": 295,
              "score": 0.9331,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 43,
              "n_models": 89,
              "score": 0.432,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 12,
              "n_models": 113,
              "score": 0.8596,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 16,
              "n_models": 70,
              "score": 0.7073,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: xhigh)",
              "rank": 17,
              "n_models": 83,
              "score": 0.74,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 14,
              "n_models": 274,
              "score": 0.9972,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 11,
              "n_models": 170,
              "score": 0.3,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 17,
              "n_models": 193,
              "score": 0.839,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 12,
            "share": 0.99,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 13,
              "n_models": 42,
              "score": 0.247,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 10,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 43,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 32,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 40,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gpt-5-6-luna",
      "name": "GPT-5.6 Luna",
      "vendor": "OpenAI",
      "released": "2026-07-09",
      "context_window": 1000000,
      "price_input": 0.2,
      "price_output": 1.2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": 91.6,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "coding",
        "agentic",
        "speed"
      ],
      "strengths": [
        "Cheapest 5.6 tier at $0.20/$1.20 (cut 80% from launch price of $1/$6)",
        "Still competitive: Terminal-Bench 2.1 84.7%, Coding Agent Index 74.6",
        "1M context"
      ],
      "weaknesses": [
        "Weak long-context recall (MRCR 41.3% vs Sol's 91.5%) — avoid for large-doc tasks",
        "No published GPQA/AIME/SWE-bench Verified/MMLU-Pro"
      ],
      "verdict": "Use for high-volume or budget coding/agent work where you don't need deep long-context recall.",
      "confidence": "medium",
      "sources": [
        "https://developers.openai.com/api/docs/pricing",
        "https://www.vellum.ai/blog/gpt-5-6-benchmarks-explained",
        "https://artificialanalysis.ai/articles/gpt-5-6-has-landed",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/text",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 79,
      "coding_basis": "vendor coding-agent index ~74.6; no third-party yet",
      "coding_confidence": "low",
      "use_well": [
        "Good for high-volume, budget coding and agent work at $0.20/$1.20.",
        "Avoid it for long documents — its long-context recall (41.3%) is far below Sol's 91.5%."
      ],
      "price_checked": {
        "url": "https://developers.openai.com/api/docs/pricing",
        "date": "2026-07-30",
        "input": 0.2,
        "output": 1.2,
        "aggregator_note": "80% price cut announced 2026-07-30 (was $1/$6); confirmed live on OpenAI's own pricing page."
      },
      "best_for_line": "Cheapest OpenAI model · 1M ctx · $0.2/$1.2",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "openai/gpt-5.6-luna",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://developers.openai.com/api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 79,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 83,
          "basis": [
            "coding_score",
            "context_window",
            "effort_ladders"
          ]
        },
        "bulk": {
          "score": 71,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 92,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 87,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 76,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 79,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 84,
          "basis": [
            "coding_score",
            "best_for:speed"
          ]
        },
        "exec-summaries": {
          "score": 92,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.07,
          "rank": 20,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 70,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 28,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": 9,
          "usage_share": 1.08,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 80,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 19,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 65,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 38,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 21,
              "n_models": 26,
              "score": 0.359,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 39,
              "n_models": 140,
              "score": 0.5359,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 16,
              "n_models": 125,
              "score": 1667.4,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 35,
              "n_models": 153,
              "score": 0.6086,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 10,
              "n_models": 26,
              "score": 0.6719,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 70,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 39,
              "n_models": 66,
              "score": 4094.712,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 42,
              "n_models": 56,
              "score": 0.43,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 28,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 15,
            "share": 1.08,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 10,
            "share": 0.76,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 80,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 29,
              "n_models": 295,
              "score": 0.916,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 49,
              "n_models": 89,
              "score": 0.41,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2 (best of 3 rows)",
              "rank": 17,
              "n_models": 113,
              "score": 0.8211,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 17,
              "n_models": 70,
              "score": 0.6098,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 22,
              "n_models": 83,
              "score": 0.6,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 27,
              "n_models": 274,
              "score": 0.9833,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 29,
              "n_models": 170,
              "score": 0.2057,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 49,
              "n_models": 193,
              "score": 0.5958,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 21,
              "n_models": 42,
              "score": 0.227,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 19,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 65,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 38,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 42,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gemini-3-5-flash",
      "name": "Gemini 3.5 Flash",
      "vendor": "Google",
      "released": "2026-05-19",
      "context_window": 1000000,
      "price_input": 1.5,
      "price_output": 9,
      "speed_tps": 278,
      "benchmarks": {
        "swe_bench": 79.3,
        "gpqa": 92.8,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "speed",
        "coding",
        "reasoning",
        "long-context",
        "vision"
      ],
      "strengths": [
        "Near-Pro reasoning (GPQA 92.2 per Artificial Analysis) at Flash speed",
        "Fast (~278 output tokens/sec); 1M context",
        "High spot on the AA Intelligence Index for its price class"
      ],
      "weaknesses": [
        "Output price ($9) is high for a Flash tier",
        "Only SWE-bench Pro (55.1) published, not SWE-bench Verified",
        "Audio input billed higher"
      ],
      "verdict": "Use when you want most of Pro's smarts but need it fast and cheaper for high-volume chat, coding, or long-context work.",
      "confidence": "medium",
      "sources": [
        "https://ai.google.dev/gemini-api/docs/pricing",
        "https://o-mega.ai/articles/gemini-3-5-flash-benchmarks-cost-and-guide",
        "https://epoch.ai/benchmarks/swe-bench-verified",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 79,
      "coding_basis": "SWE-bench Verified 79.3% (Epoch AI independent run, high reasoning effort)",
      "coding_confidence": "high",
      "use_well": [
        "Most of Pro's smarts at speed (~278 tok/s) — good for high-volume chat and coding where latency matters.",
        "$9/M output is high for a 'fast/cheap' tier — for true bulk work drop to Flash-Lite or a cheap open model."
      ],
      "price_checked": {
        "url": "https://ai.google.dev/gemini-api/docs/pricing",
        "date": "2026-07-29",
        "input": 1.5,
        "output": 9,
        "aggregator_note": "OpenRouter shows ~$2.50 — not Google's published rate."
      },
      "best_for_line": "Google model · 1M ctx · $1.5/$9",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "google/gemini-3.5-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://ai.google.dev/gemini-api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 79,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 78,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 50,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 88,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 60,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 67,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": 84,
          "basis": [
            "coding_score",
            "best_for:speed"
          ]
        },
        "exec-summaries": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "o-mega.ai's benchmark writeup (checked 2026-09-07), citing Artificial Analysis data, states Gemini 3.5 Flash surpasses Gemini 3.1 Pro on the Terminal-Bench 2.1 coding benchmark.",
              "source_url": "https://o-mega.ai/articles/gemini-3-5-flash-benchmarks-cost-and-guide",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Gemini 3.5 Flash surpasses Gemini 3.1 Pro on Terminal-Bench 2.1 (coding), GDPval-AA (real-world agentic evaluation), MCP Atlas (scaled tool use), and Finance Agent v2."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "o-mega.ai's writeup (checked 2026-09-07) states Gemini 3.5 Flash's 1M-token context window matches the prior Flash generation, half the 2M window it cites for Gemini 3.1 Pro.",
              "source_url": "https://o-mega.ai/articles/gemini-3-5-flash-benchmarks-cost-and-guide",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "The 1M token context window is the same as the previous Flash generation but half the size of Gemini 3.1 Pro's 2M window."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.03,
          "rank": 104,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 52,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 41,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 16,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 17,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 23,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 14,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SWE-bench Verified (Epoch mirror)",
              "rank": 3,
              "n_models": 33,
              "score": 0.7934,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/swe-bench-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 38,
              "n_models": 140,
              "score": 0.5394,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 64,
              "n_models": 125,
              "score": 911.02,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 33,
              "n_models": 153,
              "score": 0.6264,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: medium)",
              "rank": 23,
              "n_models": 26,
              "score": 0.3739,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 52,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 30,
              "n_models": 66,
              "score": 5396.4183,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 52,
              "n_models": 56,
              "score": 0.275,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 41,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 64,
            "share": 0.17,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 16,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 22,
              "n_models": 295,
              "score": 0.928,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 12,
              "n_models": 89,
              "score": 0.662,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 40,
              "n_models": 113,
              "score": 0.6281,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 43,
              "n_models": 70,
              "score": 0.2683,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: high)",
              "rank": 44,
              "n_models": 83,
              "score": 0.31,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 43,
              "n_models": 274,
              "score": 0.9556,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 50,
              "n_models": 170,
              "score": 0.1314,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 28,
              "n_models": 193,
              "score": 0.7208,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 31,
              "n_models": 42,
              "score": 0.14,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 17,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 23,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 14,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 48,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gemini-2-5-flash-lite",
      "name": "Gemini 2.5 Flash-Lite",
      "vendor": "Google",
      "released": "2025",
      "context_window": 1000000,
      "price_input": 0.1,
      "price_output": 0.4,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "speed"
      ],
      "strengths": [
        "Cheapest Gemini ($0.10 in / $0.40 out)",
        "1M-token context",
        "Great for massive-scale simple tasks like classification and extraction"
      ],
      "weaknesses": [
        "Lowest capability tier — not for reasoning or complex coding",
        "Older 2.5 generation",
        "Audio input billed higher"
      ],
      "verdict": "The rock-bottom-cost Gemini for huge-volume, low-complexity jobs.",
      "confidence": "high",
      "sources": [
        "https://ai.google.dev/gemini-api/docs/pricing",
        "https://artificialanalysis.ai/models/gemini-2-5-flash-lite",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 52,
      "coding_basis": "community consensus (budget tier; no coding benchmark)",
      "coding_confidence": "low",
      "use_well": [
        "The rock-bottom Gemini for massive simple jobs — classification, extraction, cleanup at $0.10/$0.40.",
        "Don't use it for reasoning or complex code — it's the lowest capability tier on the board."
      ],
      "best_for_line": "Cheapest Google model · 1M ctx · $0.1/$0.4",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "google/gemini-2.5-flash-lite",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://ai.google.dev/gemini-api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 52,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 59,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 57,
          "basis": [
            "coding_score",
            "best_for:speed"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.17,
          "rank": 54,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 25,
            "share": 0.73,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "grok-4-5",
      "name": "Grok 4.5",
      "vendor": "xAI",
      "released": "2026-07-08",
      "context_window": 500000,
      "price_input": 2,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": 93.4,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning"
      ],
      "strengths": [
        "xAI's newest flagship, pitched as 'Opus-class'; very token-efficient (~4x fewer output tokens than Opus 4.8)",
        "Cheap output ($6/M) vs frontier peers; strong agentic coding (SWE-bench Pro 64.7%, Terminal-Bench 2.1 83.3%)",
        "75% cache discount ($0.50 cached input); bundled into Cursor"
      ],
      "weaknesses": [
        "xAI did NOT publish classic academic benchmarks (GPQA/AIME/MMLU/SWE-bench Verified) at launch",
        "500k context is smaller than Gemini/Claude's 1M",
        "Not available in EU at launch; preview-era figures may shift"
      ],
      "verdict": "The cheap frontier-class coder: reach for it when you want strong agentic coding at low, token-efficient cost.",
      "confidence": "medium",
      "sources": [
        "https://docs.x.ai/docs/models",
        "https://x.ai/news/grok-4-5",
        "https://apidog.com/blog/grok-4-5-benchmarks/",
        "https://techcrunch.com/2026/07/08/spacexai-releases-grok-4-5-which-elon-describes-as-an-opus-class-model/",
        "https://epoch.ai/benchmarks/gpqa-diamond",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/search",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": 86,
      "coding_basis": "AA Coding Index 72.5 + LMArena Code Elo #4 (1566)",
      "coding_confidence": "medium",
      "use_well": [
        "Cheap frontier-class coding — ~4x fewer output tokens than Opus-class peers means real costs run below the sticker price.",
        "The 75% cache discount ($0.50 cached input) rewards repeated-context workloads like agents and long sessions.",
        "No classic benchmarks published and not in the EU at launch — verify on your own tasks first."
      ],
      "best_for_line": "Cheapest xAI model · 500K ctx · $2/$6",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "x-ai/grok-4.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://docs.x.ai/docs/models",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 86,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 72,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": 55,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "writing": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        },
        "research": {
          "score": 77,
          "basis": [
            "benchmarks.gpqa",
            "context_window"
          ]
        },
        "extraction": {
          "score": 64,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "chat": {
          "score": 70,
          "basis": [
            "benchmarks.gpqa",
            "price_output"
          ]
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 86,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": 93,
          "basis": [
            "benchmarks.gpqa"
          ]
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.03,
          "rank": 109,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 22,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 13,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 32,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 8,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 21,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 38,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 18,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": true,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 31,
              "n_models": 140,
              "score": 0.5498,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 33,
              "n_models": 125,
              "score": 1309.28,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 68,
              "n_models": 153,
              "score": 0.4643,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: high)",
              "rank": 18,
              "n_models": 26,
              "score": 0.5376,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 22,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 40,
              "n_models": 66,
              "score": 3887.426,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 19,
              "n_models": 56,
              "score": 0.562,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 13,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 105,
            "share": 0.05,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 32,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 17,
              "n_models": 295,
              "score": 0.9343,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 32,
              "n_models": 89,
              "score": 0.483,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 46,
              "n_models": 113,
              "score": 0.5719,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 46,
              "n_models": 70,
              "score": 0.2439,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: high)",
              "rank": 43,
              "n_models": 83,
              "score": 0.31,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 33,
              "n_models": 274,
              "score": 0.9778,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 43,
              "n_models": 170,
              "score": 0.1543,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 57,
              "n_models": 193,
              "score": 0.5264,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Search Arena (34 models, 1,110,523 votes) — grounded web search, closest proxy; Arena has no dedicated research/analysis category",
            "rank": 8,
            "n_models": 34,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/search"
          }
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 33,
              "n_models": 42,
              "score": 0.14,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 21,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 38,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 18,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 33,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "deepseek-v4-flash",
      "name": "DeepSeek V4-Flash",
      "vendor": "DeepSeek",
      "released": "2026-04-24",
      "context_window": 1000000,
      "price_input": 0.44,
      "price_output": 1.32,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "long-context",
        "coding"
      ],
      "strengths": [
        "Rock-bottom price for a frontier-lineage model ($0.14/$0.28)",
        "1M-token context",
        "Cache-hit input near-free (~$0.0028/M) for repeated-prefix / agent workloads"
      ],
      "weaknesses": [
        "The 'flash' tier trades some reasoning depth vs V4-Pro",
        "No cleanly-sourced benchmarks for the flash tier specifically"
      ],
      "verdict": "Reach for it to process huge volumes of text or code cheaply when you don't need the absolute top reasoning tier.",
      "confidence": "medium",
      "sources": [
        "https://api-docs.deepseek.com/quick_start/pricing/",
        "https://raw.githubusercontent.com/BerriAI/litellm/main/model_prices_and_context_window.json",
        "https://api-docs.deepseek.com/quick_start/pricing",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://www.lindy.ai/blog/migrating-from-claude-to-deepseek",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 72,
      "coding_basis": "estimated just below V4-Pro (cheaper tier); no direct benchmark",
      "coding_confidence": "low",
      "use_well": [
        "The bulk-work machine: $0.14/$0.28 with near-free cache hits — process huge volumes of text or code for pennies.",
        "Step up to V4-Pro the moment reasoning depth starts to matter."
      ],
      "best_for_line": "DeepSeek model · 1M ctx · $0.44/$1.32",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://api-docs.deepseek.com/quick_start/pricing/",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 72,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 73,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 72,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "agents": {
          "claims": [
            {
              "sentence": "Lindy's own migration write-up describes moving its managed-agent traffic (previously Claude/Sonnet- and Gemini-backed) to DeepSeek V4-Flash, checked 2026-09-06.",
              "source_url": "https://www.lindy.ai/blog/migrating-from-claude-to-deepseek",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "to DeepSeek v4 Flash on Atlas Cloud"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "Lindy reports that on the agent traffic it migrated to DeepSeek V4-Flash, inference costs fell by about 90%, checked 2026-09-06.",
              "source_url": "https://www.lindy.ai/blog/migrating-from-claude-to-deepseek",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "inference costs fell by about 90%"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.53,
          "rank": 17,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 90,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 24,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": 3,
          "usage_share": 3.67,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 79,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 85,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 3 rows)",
              "rank": 73,
              "n_models": 140,
              "score": 0.4491,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror) (best of 2 rows)",
              "rank": 87,
              "n_models": 125,
              "score": 678.2,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 73,
              "n_models": 153,
              "score": 0.4563,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 90,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [],
          "chosen": {
            "rank": 14,
            "share": 0.06,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 24,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 6,
            "share": 3.67,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 13,
            "share": 0.26,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 79,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 66,
              "n_models": 170,
              "score": 0.0714,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 7,
            "share": 1.71,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 85,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 68,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "qwen3-max",
      "name": "Qwen3-Max",
      "vendor": "Alibaba (Qwen)",
      "released": null,
      "context_window": 256000,
      "price_input": 1.2,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "coding",
        "agentic"
      ],
      "strengths": [
        "Qwen's flagship general model; top-tier on several coding/agent benchmarks per Qwen",
        "256K context; batch API halves the price",
        "Large open ecosystem"
      ],
      "weaknesses": [
        "Newer Max releases moving closed-weight",
        "Input price tiers up to $3/M past 128K tokens",
        "No clean primary SWE-bench Verified number for this SKU"
      ],
      "verdict": "Qwen's do-everything flagship — pick it for top Qwen quality across reasoning and code when you can pay flagship rates.",
      "confidence": "medium",
      "sources": [
        "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
        "https://artificialanalysis.ai/articles/qwen3-max-thinking-everything-you-need-to-know",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 74,
      "coding_basis": "LiveCodeBench 76.7% (Qwen3 Max)",
      "coding_confidence": "medium",
      "use_well": [
        "Qwen's flagship for reasoning + code; the batch API halves the price for anything that isn't time-critical.",
        "Input pricing tiers up past 128K tokens — keep prompts under the tier break when cost matters."
      ],
      "price_checked": {
        "url": "https://www.alibabacloud.com/help/en/model-studio/billing-for-model-studio",
        "date": "2026-07-29",
        "input": 1.2,
        "output": 6,
        "aggregator_note": "OpenRouter shows ~$3.90 — matches no Alibaba tier. Ours is the 0–32K Singapore tier."
      },
      "best_for_line": "Alibaba (Qwen) model · 256K ctx · $1.2/$6",
      "availability": {
        "direct_api": true,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://www.alibabacloud.com/help/en/model-studio/billing-for-model-studio"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 74,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 58,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 74,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 385,
          "as_of": "2026-10-15",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 94,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 106,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 92,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 110,
              "n_models": 125,
              "score": 370.45,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 94,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 60,
              "n_models": 66,
              "score": 71.565,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 106,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 150,
              "n_models": 295,
              "score": 0.726,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 31,
              "n_models": 89,
              "score": 0.4875,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 89,
              "n_models": 113,
              "score": 0.1895,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 118,
              "n_models": 274,
              "score": 0.7333,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 92,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "qwen3-coder-plus",
      "name": "Qwen3-Coder-Plus",
      "vendor": "Alibaba (Qwen)",
      "released": null,
      "context_window": 1000000,
      "price_input": 1,
      "price_output": 5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "long-context"
      ],
      "strengths": [
        "Hosted, tuned coding model with up to 1M context",
        "Context-caching discount available",
        "Strong Qwen coding lineage"
      ],
      "weaknesses": [
        "Tiered pricing climbs steeply past 32K tokens",
        "Output at $5/M is not cheap-bulk territory",
        "No clean primary SWE-bench figure for this exact SKU"
      ],
      "verdict": "Use for a managed Qwen coding endpoint with very long context when you don't want to self-host the open weights.",
      "confidence": "medium",
      "sources": [
        "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
        "https://tokenmix.ai/blog/qwen3-coder-plus-review-benchmark-2026",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 73,
      "coding_basis": "Qwen coder line; estimated, no clean SKU benchmark",
      "coding_confidence": "low",
      "use_well": [
        "A managed Qwen coding endpoint with up to 1M context when you don't want to self-host open weights.",
        "Tiered pricing climbs steeply past 32K tokens — trim context or costs jump."
      ],
      "price_checked": {
        "url": "https://www.alibabacloud.com/help/en/model-studio/billing-for-model-studio",
        "date": "2026-07-29",
        "input": 1,
        "output": 5,
        "aggregator_note": "OpenRouter shows ~$3.25 — matches no Alibaba tier. Ours is the 0–32K Singapore tier."
      },
      "best_for_line": "Alibaba (Qwen) model · 1M ctx · $1/$5",
      "availability": {
        "direct_api": true,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://www.alibabacloud.com/help/en/model-studio/billing-for-model-studio"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 73,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 74,
          "basis": [
            "coding_score",
            "context_window"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 73,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "TokenMix's review (checked 2026-09-07) describes Qwen3-Coder-Plus as Alibaba's dedicated coding model, fine-tuned on 18 trillion code tokens across 300+ programming languages, separate from general Qwen3-Max.",
              "source_url": "https://tokenmix.ai/blog/qwen3-coder-plus-review-benchmark-2026",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Qwen3-Coder-Plus is Alibaba's dedicated coding model — separate from general Qwen3-Max, fine-tuned on 18 trillion code tokens across 300+ programming languages."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-16"
        },
        "agents": {
          "claims": [
            {
              "sentence": "TokenMix's review (checked 2026-09-07) states Qwen3-Coder-Plus ships optimized for function calling and tool use, relevant to agent-style coding workflows.",
              "source_url": "https://tokenmix.ai/blog/qwen3-coder-plus-review-benchmark-2026",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Qwen3-Coder-Plus ships optimized for function calling and tool use."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 362,
          "as_of": "2026-10-15",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 106,
              "n_models": 125,
              "score": 456.5,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "qwen-turbo",
      "name": "Qwen-Turbo",
      "vendor": "Alibaba (Qwen)",
      "released": null,
      "context_window": null,
      "price_input": 0.05,
      "price_output": 0.2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "speed"
      ],
      "strengths": [
        "Among the cheapest hosted models anywhere ($0.05/M input)",
        "Thinking mode available at $0.5/M output"
      ],
      "weaknesses": [
        "Small/fast model — not for hard reasoning or agentic coding",
        "Context window not clearly documented in the pricing table"
      ],
      "verdict": "Use for massive-volume simple tasks — classification, tagging, extraction, cleanup — where cost per token matters most.",
      "confidence": "medium",
      "sources": [
        "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
        "https://openrouter.ai/qwen/qwen-turbo",
        "https://electroiq.com/stats/qwen-ai-statistics/"
      ],
      "coding_score": 48,
      "coding_basis": "community consensus (small/fast tier)",
      "coding_confidence": "low",
      "use_well": [
        "Among the cheapest hosted models anywhere — massive-volume tagging and cleanup at $0.05/M input.",
        "Its thinking mode ($0.5/M output) adds cheap reasoning for slightly harder bulk work."
      ],
      "best_for_line": "Alibaba (Qwen) model · $0.05/$0.2",
      "availability": {
        "direct_api": true,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://www.alibabacloud.com/help/en/model-studio/model-pricing"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 48,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 48,
          "basis": [
            "coding_score"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 53,
          "basis": [
            "coding_score",
            "best_for:speed"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "bulk": {
          "claims": [
            {
              "sentence": "Alibaba Cloud's Model Studio pricing page (checked 2026-09-07) prices Qwen-Turbo's international endpoint at $0.05 per million input tokens and $0.2 per million output tokens, and notes a discount for batch calls, which fits high-volume low-cost workloads.",
              "source_url": "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "50% batch inference discount"
            },
            {
              "sentence": "OpenRouter's model listing for Qwen-Turbo (checked 2026-09-07) describes it as a 1M-context model built for fast, low-cost handling of simple tasks.",
              "source_url": "https://openrouter.ai/qwen/qwen-turbo",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Qwen-Turbo, based on Qwen2.5, is a 1M context model that provides fast speed and low cost, suitable for simple tasks."
            },
            {
              "sentence": "Electroiq's Qwen statistics page (checked 2026-09-07) reports that more than 90,000 enterprises use the Qwen model family, of which Qwen-Turbo is the cheapest tier, as of 2025.",
              "source_url": "https://electroiq.com/stats/qwen-ai-statistics/",
              "tier": "usage",
              "date": "2026-09-07",
              "quote": "more than 90,000 enterprises use the Qwen model family"
            }
          ],
          "reconciliation": "Alibaba's Model Studio page (checked 2026-09-07) marks Qwen-Turbo as no longer updated and points users to Qwen-Flash, while the same page and OpenRouter's listing (checked 2026-09-07) still price and serve Qwen-Turbo as an active option.",
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": null
      },
      "status": "ga",
      "adoption": "unknown",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 244,
              "n_models": 295,
              "score": 0.4179,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 239,
              "n_models": 274,
              "score": 0.0611,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "mistral-medium-3-1",
      "name": "Mistral Medium 3.1",
      "vendor": "Mistral AI",
      "released": null,
      "context_window": null,
      "price_input": 0.4,
      "price_output": 2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "cheap-bulk",
        "writing"
      ],
      "strengths": [
        "Strong price-to-quality mid-tier from a European (EU-hosted) vendor",
        "Good all-round coding and text model"
      ],
      "weaknesses": [
        "Not open-weight (Medium tier is commercial)",
        "No cleanly-sourced benchmark for this version"
      ],
      "verdict": "A solid EU-based mid-tier — reach for it when data residency or a European provider matters and you want balanced cost/quality.",
      "confidence": "medium",
      "sources": [
        "https://mistral.ai/pricing/",
        "https://www.cloudzero.com/blog/mistral-api-pricing/",
        "https://openrouter.ai/mistralai/mistral-medium-3.1",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": 62,
      "coding_basis": "community consensus (EU mid-tier; no coding benchmark surfaced)",
      "coding_confidence": "low",
      "use_well": [
        "Reach for it when EU hosting or data residency matters and you want balanced cost/quality.",
        "No cleanly-sourced benchmarks for this version — validate on your own tasks before betting volume."
      ],
      "task_copy": {
        "writing": {
          "verdict": "The budget pick of the three the data tags for prose — capable everyday writing at $2/1M out, EU-hosted if residency matters.",
          "tips": [
            "Best for short-to-mid drafts, emails, and summaries; keep instructions tight for anything long-form.",
            "It's cheap enough to iterate freely — generate variants and pick, rather than perfecting one draft."
          ]
        }
      },
      "best_for_line": "Mistral AI model · $0.4/$2",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "mistralai/mistral-medium-3.1",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://mistral.ai/pricing/",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": 62,
          "basis": [
            "coding_score"
          ]
        },
        "agents": {
          "score": 62,
          "basis": [
            "coding_score"
          ]
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": 62,
          "basis": [
            "coding_score"
          ]
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter's model page for Mistral Medium 3.1 (checked 2026-09-07) states it offers competitive accuracy relative to larger models including Claude Sonnet 3.5/3.7, Llama 4 Maverick, and Command R+, and lists coding among its strong domains.",
              "source_url": "https://openrouter.ai/mistralai/mistral-medium-3.1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "offers competitive accuracy relative to larger models like Claude Sonnet 3.5/3.7, Llama 4 Maverick, and Command R+"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "OpenRouter's pricing FAQ for Mistral Medium 3.1 (checked 2026-09-07) lists the model at $0.40 per million input tokens and $2.00 per million output tokens, with cache-read priced separately at $0.04 per million tokens.",
              "source_url": "https://openrouter.ai/mistralai/mistral-medium-3.1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Mistral Medium 3.1 costs $0.40/M input tokens and $2.00/M output tokens, with separate rates for Cache Read at $0.04/M tokens."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "writing": {
          "claims": [
            {
              "sentence": "OpenRouter's model page for Mistral Medium 3.1 (checked 2026-09-07) describes it as combining strong reasoning and multimodal performance with substantially lower cost than larger models, which supports high-volume writing use.",
              "source_url": "https://openrouter.ai/mistralai/mistral-medium-3.1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It balances state-of-the-art reasoning and multimodal performance with 8× lower cost compared to traditional large models"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-27"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 227,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 242,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gemini-3-6-flash",
      "name": "Gemini 3.6 Flash",
      "vendor": "Google",
      "released": null,
      "context_window": null,
      "price_input": 0.75,
      "price_output": 3.75,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "speed",
        "cheap-bulk"
      ],
      "strengths": [
        "Priced below Gemini 3.5 Flash on output ($7.50 vs $9.00 per M) on Google’s own pricing page"
      ],
      "weaknesses": [
        "No published benchmark figures yet — every score on this row is blank on purpose",
        "Context window not stated on the pricing page; left blank rather than guessed"
      ],
      "verdict": "Listed because it exists and its list price is published. Figures pending — nothing here is a recommendation yet.",
      "confidence": "low",
      "sources": [
        "https://ai.google.dev/gemini-api/docs/pricing"
      ],
      "coding_score": null,
      "coding_basis": "No independently published coding score for this model yet.",
      "coding_confidence": "low",
      "use_well": [
        "Cheaper per output token than Gemini 3.5 Flash. Until benchmarks publish, treat that as the only known difference."
      ],
      "price_checked": {
        "url": "https://ai.google.dev/gemini-api/docs/pricing",
        "date": "2026-07-29",
        "input": 1.5,
        "output": 7.5
      },
      "best_for_line": "Google model · $0.75/$3.75",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "google/gemini-3.6-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://ai.google.dev/gemini-api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.06,
          "rank": 81,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 33,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 42,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 14,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 21,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 13,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 42,
              "n_models": 140,
              "score": 0.5336,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 85,
              "n_models": 125,
              "score": 715.52,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 45,
              "n_models": 153,
              "score": 0.561,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: high)",
              "rank": 21,
              "n_models": 26,
              "score": 0.4668,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 33,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 36,
              "n_models": 56,
              "score": 0.469,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 42,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 60,
            "share": 0.2,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 14,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 11,
              "n_models": 295,
              "score": 0.9413,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 13,
              "n_models": 89,
              "score": 0.662,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 43,
              "n_models": 113,
              "score": 0.5895,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 48,
              "n_models": 70,
              "score": 0.2195,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: high)",
              "rank": 41,
              "n_models": 83,
              "score": 0.36,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 50,
              "n_models": 274,
              "score": 0.9417,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 56,
              "n_models": 170,
              "score": 0.1057,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 48,
              "n_models": 193,
              "score": 0.6042,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 32,
              "n_models": 42,
              "score": 0.14,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 21,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 13,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 37,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "glm-5-2",
      "name": "GLM-5.2",
      "vendor": "Z.ai (Zhipu)",
      "released": null,
      "context_window": null,
      "price_input": 1.4,
      "price_output": 4.4,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "reasoning"
      ],
      "strengths": [
        "Open-weights, and priced well under the US frontier tier on Z.ai’s own published rates"
      ],
      "weaknesses": [
        "No independently run benchmark figures in our data yet — blanks are deliberate",
        "Vendor and third-party coding claims exist but none are independently verified here"
      ],
      "verdict": "Listed because it exists and its list price is published. Figures pending — nothing here is a recommendation yet.",
      "confidence": "low",
      "sources": [
        "https://docs.z.ai/guides/overview/pricing",
        "https://raw.githubusercontent.com/BerriAI/litellm/main/model_prices_and_context_window.json",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": null,
      "coding_basis": "No independently published coding score for this model yet.",
      "coding_confidence": "low",
      "use_well": [
        "Open-weights, so self-hosting is an option the closed frontier models don’t offer. Verify quality yourself until independent figures land."
      ],
      "price_checked": {
        "url": "https://docs.z.ai/guides/overview/pricing",
        "date": "2026-07-29",
        "input": 1.4,
        "output": 4.4
      },
      "best_for_line": "Z.ai (Zhipu) model · $1.4/$4.4",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "z-ai/glm-5.2",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://docs.z.ai/guides/overview/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.51,
          "rank": 29,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 48,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 12,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 31,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": 9,
          "usage_share": 2.37,
          "arena_rank": 37,
          "expert_default": null,
          "families": 2
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 4,
          "usage_share": 5.98,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SWE-bench Verified (Epoch mirror)",
              "rank": 5,
              "n_models": 33,
              "score": 0.787,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/swe-bench-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 50,
              "n_models": 140,
              "score": 0.5116,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror) (best of 2 rows)",
              "rank": 54,
              "n_models": 125,
              "score": 1046.95,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 23,
              "n_models": 153,
              "score": 0.7012,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 22,
              "n_models": 26,
              "score": 0.4378,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 11,
            "share": 1.6,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 48,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 14,
              "n_models": 66,
              "score": 8313.7783,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 39,
              "n_models": 56,
              "score": 0.452,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 12,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 27,
            "share": 0.69,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 31,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 3 rows)",
              "rank": 27,
              "n_models": 295,
              "score": 0.9186,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 58,
              "n_models": 89,
              "score": 0.342,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2 (best of 3 rows)",
              "rank": 42,
              "n_models": 113,
              "score": 0.5921,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 37,
              "n_models": 70,
              "score": 0.2927,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: max)",
              "rank": 42,
              "n_models": 83,
              "score": 0.35,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 3 rows)",
              "rank": 85,
              "n_models": 274,
              "score": 0.8639,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 27,
              "n_models": 170,
              "score": 0.2086,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 78,
              "n_models": 193,
              "score": 0.2278,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 4,
            "share": 2.71,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 37,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 4,
            "share": 5.98,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 21,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gemini-3-7-flash",
      "name": "Gemini 3.7 Flash",
      "vendor": "Google",
      "released": "2026-08-13",
      "context_window": 1048576,
      "price_input": 0.75,
      "price_output": 3.75,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/google/gemini-3.7-flash",
        "https://ai.google.dev/gemini-api/docs/pricing",
        "https://openrouter.ai/api/v1/models",
        "https://ai.google.dev/gemini-api/docs/latest-model",
        "https://deepmind.google/models/gemini/flash/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Use low thinking effort for latency-critical work like real-time chat, fast drafts, or incident-response pipelines.",
        "Leave it on medium (the default) for most coding and agentic tasks — it gives the best first-pass accuracy for the cost.",
        "Reserve high thinking effort for the hardest reasoning, math, and multi-step agent tasks; it burns more tokens and time.",
        "MINIMAL thinking level is not supported on 3.7 Flash — only low, medium, and high — so don't set it or the API will reject the call."
      ],
      "task_copy": {},
      "auto_added": "2026-08-16",
      "best_for_line": "Google model · 1M ctx · $0.75/$3.75",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "google/gemini-3.7-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://ai.google.dev/gemini-api/docs/pricing",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.21,
          "rank": 49,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 23,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 35,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 3,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 11,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 3 rows)",
              "rank": 10,
              "n_models": 140,
              "score": 0.5984,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 65,
              "n_models": 125,
              "score": 904.3,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: medium)",
              "rank": 12,
              "n_models": 26,
              "score": 0.6549,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 23,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 7,
              "n_models": 56,
              "score": 0.678,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 35,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 33,
            "share": 0.58,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 3,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 6,
              "n_models": 295,
              "score": 0.9482,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 10,
              "n_models": 89,
              "score": 0.692,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 26,
              "n_models": 113,
              "score": 0.7158,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 28,
              "n_models": 70,
              "score": 0.3659,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 23,
              "n_models": 83,
              "score": 0.58,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 34,
              "n_models": 274,
              "score": 0.9722,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 3 rows)",
              "rank": 47,
              "n_models": 170,
              "score": 0.1429,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 13,
              "n_models": 193,
              "score": 0.8458,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 16,
              "n_models": 42,
              "score": 0.238,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 11,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 22,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "grok-4-6",
      "name": "Grok 4.6",
      "vendor": "xAI",
      "released": "2026-08-12",
      "context_window": 500000,
      "price_input": 2,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "agentic",
        "coding",
        "vision",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/x-ai/grok-4.6",
        "https://x.ai/news/grok-4-6",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://docs.x.ai/developers/models/grok-4.6"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Grok 4.6 supports four reasoning efforts (low, medium, high, xhigh); reserve high or xhigh for genuinely hard multi-step problems since deeper thinking costs more tokens and latency.",
        "Cache repeated context: xAI prices cached input at $0.50 per million tokens versus $2 standard for prompts under 200K tokens, so reusing a long system prompt or codebase is much cheaper.",
        "Its 500K context window and image input make it well suited to long-document or codebase review that includes visual inputs, not just short chat turns.",
        "At $2/$6 per million input/output tokens, use a cheaper tier for simple lookups and save Grok 4.6 for long-running agentic or coding work it's built for."
      ],
      "task_copy": {},
      "auto_added": "2026-08-16",
      "best_for_line": "Priciest xAI model · 500K ctx · $2/$6",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "x-ai/grok-4.6",
        "aws_bedrock": true,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/current/us-east-1/index.json",
          "https://x.ai/news/grok-4-6"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "agents": {
          "claims": [
            {
              "sentence": "xAI's own launch post (Aug 12, 2026) says Grok 4.6 matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index, a composite of nine benchmarks, and improves on Grok 4.5's chart score of 56 with a score of 61.",
              "source_url": "https://x.ai/news/grok-4-6",
              "tier": "lab",
              "date": "2026-08-12",
              "quote": "It matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index, which is a composite score of nine benchmarks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.05,
          "rank": 92,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 45,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 19,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 26,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 49,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 10,
          "expert_default": true,
          "families": 2
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Extra High)",
              "rank": 17,
              "n_models": 26,
              "score": 0.414,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 4 rows)",
              "rank": 24,
              "n_models": 140,
              "score": 0.5648,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 26,
              "n_models": 125,
              "score": 1508.08,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 26,
              "n_models": 153,
              "score": 0.6729,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: medium)",
              "rank": 9,
              "n_models": 26,
              "score": 0.6748,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 45,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 13,
              "n_models": 66,
              "score": 9047.0325,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 10,
              "n_models": 56,
              "score": 0.653,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 19,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 82,
            "share": 0.1,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 26,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 2 rows)",
              "rank": 12,
              "n_models": 295,
              "score": 0.94,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified (best of 2 rows)",
              "rank": 30,
              "n_models": 89,
              "score": 0.493,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 34,
              "n_models": 113,
              "score": 0.6596,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 32,
              "n_models": 70,
              "score": 0.3171,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 30,
              "n_models": 83,
              "score": 0.51,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 2 rows)",
              "rank": 19,
              "n_models": 274,
              "score": 0.9917,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 4 rows)",
              "rank": 31,
              "n_models": 170,
              "score": 0.1971,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 35,
              "n_models": 193,
              "score": 0.6708,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 26,
              "n_models": 42,
              "score": 0.172,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 49,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 15,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "muse-spark-1-2",
      "name": "Muse Spark 1.2",
      "vendor": "Meta",
      "released": "2026-08-05",
      "context_window": 1048576,
      "price_input": 1.25,
      "price_output": 4.25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-spark-1.2",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/vision",
        "https://research.meta.ai/blog/introducing-muse-code-and-muse-spark-1-2"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Meta's own launch post points this model at whole-repository generation, large end-to-end projects, and auto-research, not just code completion.",
        "Pair it with Meta's Muse Code terminal agent when possible — the two were co-trained together specifically so they perform best as a pair.",
        "Its 1,048,576-token context window means you can hand it a full large repo rather than chunking files across turns."
      ],
      "task_copy": {},
      "auto_added": "2026-08-16",
      "best_for_line": "Meta model · 1M ctx · $1.25/$4.25",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-spark-1.2",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 209,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 10,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 34,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 33,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 5,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 8,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 21,
              "n_models": 140,
              "score": 0.5741,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 40,
              "n_models": 153,
              "score": 0.603,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: xhigh)",
              "rank": 16,
              "n_models": 26,
              "score": 0.5487,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 10,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 38,
              "n_models": 56,
              "score": 0.454,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 34,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 188,
            "share": 0.01,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 33,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 16,
              "n_models": 89,
              "score": 0.603,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 37,
              "n_models": 83,
              "score": 0.43,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 38,
              "n_models": 170,
              "score": 0.1771,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 29,
              "n_models": 42,
              "score": 0.16,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 5,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 8,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 39,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "muse-spark-1-1",
      "name": "Muse Spark 1.1",
      "vendor": "Meta",
      "released": "2026-07-16",
      "context_window": 1048576,
      "price_input": 1.25,
      "price_output": 4.25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "agentic",
        "coding",
        "long-context",
        "vision"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-spark-1.1",
        "https://ai.meta.com/blog/introducing-muse-spark-msl/",
        "https://ai.meta.com/static-resource/muse-spark-1-1-evaluation-report/",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Natively multimodal — it accepts image, video, PDF and audio input directly, so skip a separate OCR or transcription step for those tasks.",
        "1M-token context window suits large codebases or long documents in a single pass rather than chunking.",
        "Scores 76.0 on SWE-Bench Verified and 51.2 on SWE-Bench Pro per Meta's own evaluation report — solid for autonomous coding-agent work, not just chat.",
        "At $1.25/$4.25 per million input/output tokens, output-heavy agentic loops (tool calls, long reasoning traces) dominate cost, so budget for output volume."
      ],
      "task_copy": {},
      "auto_added": "2026-08-16",
      "best_for_line": "Meta model · 1M ctx · $1.25/$4.25",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-spark-1.1",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for Muse Spark 1.1 (checked 2026-09-07) relays Meta's claim of strong performance on real-world coding across large codebases, computer-use workflows, and visual-to-code generation.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Meta reports strong performance on real-world coding across large codebases, computer-use workflows, and visual-to-code generation."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for Muse Spark 1.1 (checked 2026-09-07) says it is built to orchestrate multi-agent workflows as either a main planning agent or a subagent, and to generalize to new tools and MCP servers without extra training.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "The model is designed to orchestrate multi-agent workflows, acting as either a main agent that plans and delegates or as a subagent"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "vision": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for Muse Spark 1.1 (checked 2026-10-03) states it accepts text, images, video and PDF documents as input with a 1M-token context window.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.1",
              "tier": "reported",
              "date": "2026-10-03",
              "quote": "It accepts text, images, video, and PDF documents and returns text, with a 1M-token context window."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-03"
        },
        "research": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for Muse Spark 1.1 (checked 2026-09-07) says it supports structured output, parallel function calling, built-in search with citations, and configurable reasoning effort, which fits research-style tasks.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It supports structured output, parallel function calling, built-in search with citations, and configurable reasoning effort."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 322,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 12,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 37,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 43,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 13,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 10,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 22,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 15,
              "n_models": 140,
              "score": 0.588,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: xhigh)",
              "rank": 19,
              "n_models": 26,
              "score": 0.5332,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 12,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 20,
              "n_models": 66,
              "score": 6520.475,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 50,
              "n_models": 56,
              "score": 0.318,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 37,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 288,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 43,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 18,
              "n_models": 89,
              "score": 0.5779,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: xhigh)",
              "rank": 39,
              "n_models": 83,
              "score": 0.39,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 45,
              "n_models": 170,
              "score": 0.1514,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 30,
              "n_models": 42,
              "score": 0.15,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Document Arena (39 models, 322,650 votes)",
            "rank": 13,
            "n_models": 39,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/document"
          }
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 10,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 22,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 35,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "agentic",
        "coding",
        "long-context",
        "speed",
        "cheap-bulk"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Thinking mode is on by default; disable it per request with chat_template_kwargs: {enable_thinking: false} when you just need a fast, non-reasoning answer.",
        "Only 5.1B of its 124B parameters activate per token, so it runs cheap relative to its benchmark scores — a good pick for high-volume or cheap-bulk coding-agent traffic.",
        "Deploy with SGLang or vLLM using tensor parallelism across 4–8 GPUs; its HiCache+Mooncake hierarchical caching cuts time-to-first-token 60–80% on long-input requests.",
        "Use the model card's recommended sampling settings (temperature 0.6, top_p 0.95, top_k 20) — its reasoning traces are tuned around them."
      ],
      "task_copy": {},
      "id": "ling-3-0-flash",
      "name": "Ling-3.0-flash",
      "vendor": "InclusionAI (Ant Group)",
      "context_window": 262144,
      "price_input": 0.06,
      "price_output": 0.18,
      "sources": [
        "https://huggingface.co/inclusionAI/Ling-3.0-flash",
        "https://blog.kilo.ai/p/announcing-ling-30-flash-free-on",
        "https://www.aimadetools.com/blog/ling-3-0-flash-complete-guide/",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "InclusionAI (Ant Group) model · 262K ctx · $0.06/$0.18",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "inclusionai/ling-3.0-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "InclusionAI's own Hugging Face model card for Ling-3.0-flash (checked 2026-09-07) states the model matches or exceeds its predecessor Ring-2.6-1T across key benchmarks while activating far fewer parameters per token.",
              "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Ling-3.0-flash matches or outperforms its predecessor across key benchmarks."
            },
            {
              "sentence": "Kilo's blog post on Ling 3.0 Flash (checked 2026-09-07) says it is built for running multi-turn coding agents and daily agentic workflows on tight token budgets.",
              "source_url": "https://blog.kilo.ai/p/announcing-ling-30-flash-free-on",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Whether you are running multi-turn coding agents or building daily agentic workflows on tight token budgets, Ling 3.0 Flash brings high capabilities at lightning speed."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "InclusionAI's Hugging Face model card (checked 2026-09-07) says Ling-3.0-flash activates only 5.1B parameters per token while providing reasoning, instruction-following, and long-context capabilities meant to power complex agentic workflows in production.",
              "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Activating only 5.1B parameters per token, it provides impressive reasoning, instruction following, and long-context capabilities to empower complex agentic workflows in production environments."
            },
            {
              "sentence": "Kilo's blog post (checked 2026-09-07) describes Ling 3.0 Flash as suited to daily agentic workflows run on tight token budgets.",
              "source_url": "https://blog.kilo.ai/p/announcing-ling-30-flash-free-on",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Whether you are running multi-turn coding agents or building daily agentic workflows on tight token budgets, Ling 3.0 Flash brings high capabilities at lightning speed."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "InclusionAI's Hugging Face model card (checked 2026-09-07) says Ling-3.0-flash is engineered for speed and compute efficiency and claims it performs well against larger competing models and its own previous-generation flagships.",
              "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Engineered for speed, compute efficiency, and production deployment, Ling-3.0-flash delivers class-defying performance against both larger SOTA competitors and previous-generation flagships."
            }
          ],
          "reconciliation": "InclusionAI's own claim of beating larger competitors is self-reported; the third-party Kilo blog places Ling 3.0 Flash alongside Gemini 3.6 Flash contextually but does not report a direct head-to-head benchmark between the two as of 2026-09-07.",
          "as_of": "2026-09-07"
        },
        "research": {
          "claims": [
            {
              "sentence": "InclusionAI's Hugging Face model card (checked 2026-09-07) credits Ling-3.0-flash with long-context capabilities meant to support complex, multi-step workflows.",
              "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Activating only 5.1B parameters per token, it provides impressive reasoning, instruction following, and long-context capabilities to empower complex agentic workflows in production environments."
            },
            {
              "sentence": "Kilo's blog post (checked 2026-09-07) says Ling 3.0 Flash ships with a native 256K context window, extendable to 1M, aimed at long-horizon task stability and deep repository reasoning.",
              "source_url": "https://blog.kilo.ai/p/announcing-ling-30-flash-free-on",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Features a native 256K context window (extendable up to 1M), perfect for long-horizon task stability and deep repository reasoning."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.06,
          "rank": 82,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 51,
            "share": 0.27,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "kat-coder-pro-v2-5",
      "name": "KAT-Coder-Pro V2.5",
      "vendor": "Kwaipilot",
      "sources": [
        "https://www.marktechpost.com/2026/07/26/kwaikat-team-releases-kat-coder-v2-5-an-agentic-coding-model-trained-on-100000-verifiable-repository-environments/",
        "https://vercel.com/ai-gateway/models/kat-coder-pro-v2.5",
        "https://arxiv.org/abs/2607.05471",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arxiv.org/html/2607.05471"
      ],
      "best_for_line": "Kwaipilot model",
      "availability": {
        "direct_api": null,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "The KAT-Coder-V2.5 technical report on arXiv (checked 2026-09-07) states the model ranks second only to Opus 4.8 on repository-level software engineering benchmarks.",
              "source_url": "https://arxiv.org/abs/2607.05471",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "ranks second only to the frontier Opus 4.8 on repository-level software engineering"
            },
            {
              "sentence": "The results table in the KAT-Coder-V2.5 technical report on arXiv (checked 2026-10-04) lists the model second on SWE-Bench Pro (65.2 vs 69.2 for Opus 4.8) and on KAT Code Bench (53.1 vs 57.3).",
              "source_url": "https://arxiv.org/html/2607.05471",
              "tier": "lab",
              "date": "2026-10-04",
              "quote": "Opus 4.8 Coding SWE-Bench Pro 65.2 58.4 62.1 58.6 69.2 KAT Code Bench 53.1 49.6 50.3 48.9 57.3"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-04"
        },
        "agents": {
          "claims": [
            {
              "sentence": "The KAT-Coder-V2.5 technical report on arXiv (checked 2026-09-07) credits the model with the highest PinchBench agentic tool-use score among the models the KwaiKAT team evaluated.",
              "source_url": "https://arxiv.org/abs/2607.05471",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "delivers the best agentic tool-use result on PinchBench"
            },
            {
              "sentence": "The KAT-Coder-V2.5 technical report on arXiv (checked 2026-10-04) says the model's PinchBench result edges out Opus 4.8 and surpasses the GLM-5 series by a wide margin.",
              "source_url": "https://arxiv.org/html/2607.05471",
              "tier": "lab",
              "date": "2026-10-04",
              "quote": "On the Claw category, KAT-Coder-V2.5 attains the best overall result on PinchBench, edging out Opus 4.8 and surpassing the GLM-5 series by a wide margin."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-04"
        }
      },
      "usage": {
        "openrouter": null
      },
      "status": "ga",
      "adoption": "unknown",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "price_input": 0.74,
      "price_output": 2.96
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "kat-coder-air-v2-5",
      "name": "KAT-Coder-Air V2.5",
      "vendor": "Kwaipilot",
      "sources": [
        "https://www.marktechpost.com/2026/07/26/kwaikat-team-releases-kat-coder-v2-5-an-agentic-coding-model-trained-on-100000-verifiable-repository-environments/",
        "https://openrouter.ai/kwaipilot/kat-coder-air-v2.5",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://developer.puter.com/ai/kwaipilot/kat-coder-air-v2.5/"
      ],
      "best_for_line": "Kwaipilot model",
      "availability": {
        "direct_api": null,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for KAT-Coder-Air V2.5 (checked 2026-09-07) says it can take an entire issue or business workflow and autonomously locate and make the needed code changes inside the actual repository.",
              "source_url": "https://openrouter.ai/kwaipilot/kat-coder-air-v2.5",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "KAT-Coder-Air V2.5 is a flagship-level Agentic Coding model that can directly hand over an entire issue or an entire business workflow to it"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for KAT-Coder-Air V2.5 (checked 2026-09-07) describes it as able to complete an entire process autonomously inside a real repository rather than a single-turn edit.",
              "source_url": "https://openrouter.ai/kwaipilot/kat-coder-air-v2.5",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "allowing it to autonomously locate and make modifications, and complete the entire process in the actual repository"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "frontend": {
          "claims": [
            {
              "sentence": "OpenRouter's model description for KAT-Coder-Air V2.5 (checked 2026-09-07) says it retains the front-end aesthetic generation capability of the prior V2 model through an integrated mixture of experts.",
              "source_url": "https://openrouter.ai/kwaipilot/kat-coder-air-v2.5",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "it seamlessly integrates multiple experts to fully retain the front-end aesthetic generation capability of V2"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "Puter's developer listing for KAT-Coder-Air V2.5 (checked 2026-09-07) prices it at $0.15 per million input tokens and $0.6 per million output tokens.",
              "source_url": "https://developer.puter.com/ai/kwaipilot/kat-coder-air-v2.5/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "KAT-Coder-Air V2.5 costs $0.15 per 1M input tokens and $0.6 per 1M output tokens."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": null
      },
      "status": "ga",
      "adoption": "unknown",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": 59.5,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "longcat-2-0",
      "name": "LongCat 2.0",
      "vendor": "Meituan",
      "context_window": 1000000,
      "sources": [
        "https://github.com/meituan-longcat/LongCat-2.0",
        "https://venturebeat.com/technology/meituan-open-sources-longcat-2-0-the-1-6t-near-frontier-agentic-coding-model-thats-been-leading-openrouter-trained-entirely-on-chinese-chips",
        "https://www.scmp.com/tech/tech-trends/article/3358854/china-debuts-biggest-ai-model-trained-local-chips-meituan-releases-longcat-20",
        "https://felloai.com/longcat-2-0/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://www.opensourceforu.com/2026/06/meituan-open-sources-longcat-2-0-under-mit-license/"
      ],
      "best_for_line": "Meituan model · 1M ctx",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meituan/longcat-2.0",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Meituan's LongCat-2.0 GitHub README (checked 2026-09-07) states the model's architecture and post-training give it strong performance on coding and agentic tasks.",
              "source_url": "https://github.com/meituan-longcat/LongCat-2.0",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "this gives LongCat-2.0 strong performance on coding and agentic tasks"
            },
            {
              "sentence": "Meituan's LongCat-2.0 GitHub README (checked 2026-10-04) lists its own SWE-bench Pro score of 59.5 for LongCat-2.0 beside 58.6 for GPT-5.5 and 69.2 for Claude Opus 4.8.",
              "source_url": "https://github.com/meituan-longcat/LongCat-2.0",
              "tier": "lab",
              "date": "2026-10-04",
              "quote": "SWE-bench Pro 59.5 54.2* 58.6* 57.3* 64.3* 69.2*"
            }
          ],
          "reconciliation": "Meituan's README reports these SWE-bench Pro scores itself (59.5 for LongCat-2.0, 58.6 for GPT-5.5, 69.2 for Claude Opus 4.8); no independent SWE-bench Pro run of LongCat-2.0 is on file as of 2026-10-04.",
          "as_of": "2026-10-04"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Meituan's LongCat-2.0 GitHub README (checked 2026-09-07) says the model is deeply integrated with harnesses such as Claude Code, OpenClaw, and Hermes for repository-level edits and automated task execution.",
              "source_url": "https://github.com/meituan-longcat/LongCat-2.0",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "delivering strong performance across code understanding, repository-level edits, automated task execution, and agentic workflows"
            },
            {
              "sentence": "Open Source For You reported (checked 2026-10-04) that Owl Alpha, the anonymous model later revealed as LongCat-2.0, processed around 10.1 trillion tokens a month before its public unveiling.",
              "source_url": "https://www.opensourceforu.com/2026/06/meituan-open-sources-longcat-2-0-under-mit-license/",
              "tier": "reported",
              "date": "2026-10-04",
              "quote": "Before its public unveiling, Owl Alpha processed around 10.1 trillion monthly tokens, averaging 559 billion tokens daily"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-04"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 214,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 252,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "vision"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Meta's own launch post names local agents and function calling, local coding, and LLM-as-a-judge evaluation as its target use cases — it's built to run fully offline on a Mac or PC with one consumer GPU.",
        "Use its 'Controllable Effort' setting to trade quality for speed rather than switching models when you just need a faster pass.",
        "Turn on speculative decoding via the DFlash drafter when you need faster generation without changing the base model."
      ],
      "task_copy": {},
      "id": "muse-glimmer-30b",
      "name": "Muse Glimmer 30B",
      "vendor": "Meta",
      "sources": [
        "https://developer.meta.com/ai/models/muse-glimmer/",
        "https://huggingface.co/meta-models/Muse-Glimmer-30B",
        "https://venturebeat.com/technology/meta-returns-to-open-source-with-muse-glimmer-an-apache-2-0-licensed-30b-parameter-ai-model-optimized-for-agents-available-now",
        "https://kingy.ai/blog/muse-glimmer-30b-benchmarks-hardware-run/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model",
        "https://dev.meta.ai/models/muse-glimmer/"
      ],
      "best_for_line": "Meta model",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-glimmer-30b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Meta's Hugging Face model card for Muse Glimmer 30B (checked 2026-09-07) lists coding agent work as an intended use case, covering writing, debugging, and resolving software engineering tasks.",
              "source_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Coding agents: Writing, debugging, and resolving real-world software engineering tasks"
            },
            {
              "sentence": "Independent review site Kingy.ai (checked 2026-09-07) noted that Meta's published SWE-Bench Pro and SWE-Bench Verified comparison table for Muse Glimmer 30B is Meta's own data, not an independently reproduced benchmark run.",
              "source_url": "https://kingy.ai/blog/muse-glimmer-30b-benchmarks-hardware-run/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "it is not a benchmark sweep, a zero-cost service or a fully reproducible open-source release"
            }
          ],
          "reconciliation": "Meta's own SWE-Bench Verified score of 76.0 for Muse Glimmer 30B is close to but below the 77.2 Meta recorded for Qwen3.6-27B in the same table, and Kingy.ai's 2026-09-07-checked review flagged that table as Meta-sourced rather than independently reproduced.",
          "as_of": "2026-09-07"
        },
        "vision": {
          "claims": [
            {
              "sentence": "Meta's Hugging Face model card (checked 2026-09-07) lists multimodal reasoning over screenshots, charts, documents, and images as an intended use case for Muse Glimmer 30B.",
              "source_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Multimodal reasoning: Interpreting screenshots, charts, documents, and images"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Meta's own developer page for Muse Glimmer 30B (checked 2026-10-16) bills it as an open model built for always-on local agents, with reliable tool-calling and persistent state across restarts.",
              "source_url": "https://dev.meta.ai/models/muse-glimmer/",
              "tier": "lab",
              "date": "2026-10-16",
              "quote": "An open model built for always-on local agents."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-16"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.01,
          "rank": 164,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 145,
            "share": 0.03,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "price_input": 0.35,
      "price_output": 1.5
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "speed",
        "cheap-bulk"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Nemotron 3.5 Lightning is NVIDIA's open 30B-parameter MoE (3B active) tuned for long-running autonomous agents — treat it as a workhorse sub-agent rather than a single high-stakes reasoning call.",
        "It is sized to run on a single H100 or DGX Spark, so self-hosting it is a realistic way to cut cost for latency-sensitive agent loops instead of paying per-token for a hosted frontier model on every sub-task.",
        "Context extends up to 1M tokens but completion output is capped at 32,768 tokens per turn — plan multi-turn continuation for long agent trajectories rather than expecting one giant response."
      ],
      "task_copy": {},
      "id": "nemotron-3-5-lightning",
      "name": "Nemotron 3.5 Lightning",
      "vendor": "NVIDIA",
      "sources": [
        "https://developer.nvidia.com/blog/nvidia-nemotron-3-5-lightning-delivers-fast-accurate-specialized-task-execution-for-long-running-agents/",
        "https://blogs.nvidia.com/blog/nemotron-lightning-switchyard-rtx-dgx/",
        "https://www.cnbc.com/2026/08/11/nvidia-releases-nemotron-3point5-lightning-open-source-ai-model-.html",
        "https://build.nvidia.com/nvidia/nemotron-3.5-lightning-30b-a3b/modelcard",
        "https://openrouter.ai/nvidia/nemotron-3.5-lightning"
      ],
      "best_for_line": "NVIDIA model",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "nvidia/nemotron-3.5-lightning",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "agents": {
          "claims": [
            {
              "sentence": "CNBC reported on August 11, 2026 that Nvidia said Nemotron 3.5 Lightning was developed particularly for agents, AI programs that run autonomously in the background, and that CrowdStrike, CodeRabbit, and Harvey have tested and customized the model.",
              "source_url": "https://www.cnbc.com/2026/08/11/nvidia-releases-nemotron-3point5-lightning-open-source-ai-model-.html",
              "tier": "reported",
              "date": "2026-08-11",
              "quote": "The chipmaker said it was developed particularly for agents, or AI programs that can run autonomously in the background."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.5,
          "rank": 30,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 46,
            "share": 0.34,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "price_input": 0.06
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Thinking is on by default and earns its cost here: Poolside's own launch post reports max thinking lifting Terminal-Bench 2.1 from 60.4% to 70.2% and DeepSWE from 16.5% to 40.4%.",
        "Reach for no-thinking mode only for quick, simple queries where speed matters more than depth — thinking mode burns roughly 4x the completion tokens (about 101k vs 23k per trajectory on SWE Multilingual).",
        "There's no user-configurable low/medium/high effort control yet, per the vendor — you get thinking on or off, not a dial.",
        "It verifies its work before declaring done, so don't pad your prompt with 'double-check your answer' instructions — they're redundant here."
      ],
      "task_copy": {},
      "id": "laguna-s-2-1",
      "name": "Laguna S 2.1",
      "vendor": "Poolside",
      "context_window": 1000000,
      "sources": [
        "https://poolside.ai/blog/introducing-laguna-s-2-1",
        "https://venturebeat.com/infrastructure/poolside-drops-laguna-s-2-1-an-open-weight-coding-model-that-beats-rivals-10x-its-size",
        "https://aitoolsreview.co.uk/insights/poolside-laguna-s-2-1",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Poolside model · 1M ctx",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "poolside/laguna-s-2.1",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Poolside's July 21, 2026 launch blog post for Laguna S 2.1 (checked 2026-09-07) states the model is a 118B total parameter Mixture-of-Experts model with 8B activated parameters per token.",
              "source_url": "https://poolside.ai/blog/introducing-laguna-s-2-1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "a 118b total parameter mixture-of-experts (moe) model with 8b activated parameters per token"
            },
            {
              "sentence": "AIToolsReview's review of Laguna S 2.1 (checked 2026-09-07) found the model winning outright against named rivals on SWE-Bench Multilingual and SWE Atlas while trailing them on other benchmarks.",
              "source_url": "https://aitoolsreview.co.uk/insights/poolside-laguna-s-2-1",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "laguna s 2.1 actually wins outright on swe-bench multilingual (78.5%) and swe atlas (46.2%)"
            }
          ],
          "reconciliation": "Poolside's own July 21, 2026 blog post describes Laguna S 2.1 as the leading agentic coding model within its own weight class, while AIToolsReview's 2026-09-07-checked review found it trailing the much larger Kimi K3 on Terminal-Bench 2.1 and Claude Fable 5 on SWE-Bench Pro.",
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Poolside's launch blog post (checked 2026-09-07) describes Laguna S 2.1 as a step forward in developing models that pursue longer-horizon work and make effective use of reasoning.",
              "source_url": "https://poolside.ai/blog/introducing-laguna-s-2-1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "a significant step forward in our development of models that pursue longer horizon work and make effective use of reasoning"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "price_input": 0.09,
      "price_output": 0.18,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.49,
          "rank": 31,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 55,
            "share": 0.26,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "vision",
        "agentic",
        "coding"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "It's a vision-language model built for multimodal agent work like Search Agent and CI Agent — reach for it when the task involves images or video, not pure text.",
        "Function calling, structured outputs, and context caching work in most regions, but fine-tuning and batch inference availability varies — check your region before relying on those.",
        "Alibaba's own docs say it gives \"a smoother vibe coding experience\" — a good fit for coding-adjacent agent loops, not just chat."
      ],
      "task_copy": {},
      "id": "qwen3-7-flash",
      "name": "Qwen3.7 Flash",
      "vendor": "Alibaba (Qwen)",
      "context_window": 991000,
      "price_input": 0.03,
      "price_output": 0.13,
      "sources": [
        "https://docs.qwencloud.com/changelog/models",
        "https://www.eesel.ai/blog/qwen-3-7-flash",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://help.aliyun.com/en/model-studio/qwen3-7-flash",
        "https://www.alibabacloud.com/help/en/model-studio/qwen3-7-flash"
      ],
      "best_for_line": "Cheapest Alibaba (Qwen) model · 991K ctx · $0.03/$0.13",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "qwen/qwen3.7-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://docs.qwencloud.com/changelog/models",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "bulk": {
          "claims": [
            {
              "sentence": "Independent review site eesel.ai (checked 2026-09-07) recommends Qwen3.7 Flash specifically for high-volume multimodal work where an occasional wrong answer is tolerable, such as tagging and classification.",
              "source_url": "https://www.eesel.ai/blog/qwen-3-7-flash",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "high-volume multimodal work where a single wrong answer is cheap and recoverable"
            },
            {
              "sentence": "Alibaba's QwenCloud changelog (checked 2026-09-07) describes Qwen3.7 Flash as delivering a comprehensive upgrade over the prior 3.6-Flash model in multimodal understanding and agent execution.",
              "source_url": "https://docs.qwencloud.com/changelog/models",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "delivers a comprehensive upgrade over 3.6-Flash in multimodal understanding and agent execution"
            }
          ],
          "reconciliation": "Alibaba's own changelog markets Qwen3.7 Flash's multimodal and agent upgrades in prose without a benchmark table, while eesel.ai's 2026-09-07-checked review independently ran Roboflow Vision Evals and placed it 22nd of 23 models overall, recommending it only for cheap, error-tolerant, high-volume classification work.",
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.22,
          "rank": 48,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 21,
            "share": 0.96,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond (best of 2 rows)",
              "rank": 102,
              "n_models": 295,
              "score": 0.8232,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2 (best of 2 rows)",
              "rank": 87,
              "n_models": 113,
              "score": 0.193,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 2 rows)",
              "rank": 81,
              "n_models": 274,
              "score": 0.8667,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "seed-2-1-turbo",
      "name": "Seed 2.1 Turbo",
      "vendor": "ByteDance",
      "context_window": 256000,
      "sources": [
        "https://seed.bytedance.com/en/seed2_1",
        "https://datanorth.ai/news/bytedance-releases-seed-2-1-pro-and-seed-2-1-turbo",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "ByteDance model · 256K ctx",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "bytedance-seed/seed-2-1-turbo",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "ByteDance's Seed 2.1 page (checked 2026-09-07) states that in code engineering the model strengthens end-to-end delivery with more reliable requirement understanding, coding, debugging, and validation.",
              "source_url": "https://seed.bytedance.com/en/seed2_1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "in code engineering, it strengthens end-to-end delivery, enabling more reliable requirement understanding, coding, debugging, and validation"
            }
          ],
          "reconciliation": "ByteDance's own Seed 2.1 page describes the Turbo variant's coding capabilities in prose, but per DataNorth's 2026-09-07-checked report ByteDance has not published parameter counts or independent benchmark scores specifically for the Seed 2.1 family.",
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "ByteDance's Seed 2.1 page (checked 2026-09-07) states the model reliably handles multi-step tasks ranging from project planning and file processing to tool use.",
              "source_url": "https://seed.bytedance.com/en/seed2_1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "from project planning and file processing to tool use"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "exec-summaries": {
          "claims": [
            {
              "sentence": "ByteDance's Seed 2.1 page (checked 2026-09-07) states the model reliably handles multi-step tasks for high-value office work and complex everyday consultations.",
              "source_url": "https://seed.bytedance.com/en/seed2_1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "for high-value office work and complex everyday consultations, it reliably handles multi-step tasks"
            },
            {
              "sentence": "DataNorth's June 24, 2026 report on the Seed 2.1 launch (checked 2026-09-07) describes Seed 2.1 Turbo as a faster, lower-cost variant built for high-frequency enterprise workloads.",
              "source_url": "https://datanorth.ai/news/bytedance-releases-seed-2-1-pro-and-seed-2-1-turbo",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "a faster, lower cost variant built for high frequency enterprise workloads"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 242,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 247,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "price_input": 0.5,
      "price_output": 2.5
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "writing"
      ],
      "strengths": [
        "Stronger narrative structure and tension in collaborative storytelling",
        "Nuanced handling of mature and darker themes"
      ],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Aion-3.0 is a multi-model system built on the GLM family, purpose-built for character-driven fiction and roleplay, not general reasoning or coding tasks.",
        "Use it when you need stronger narrative structure, tension, and nuanced handling of mature or darker themes — its stated strengths — not for short factual Q&A.",
        "Output is capped at 32K tokens per response, so continue long-form scenes across multiple calls rather than expecting one giant completion."
      ],
      "task_copy": {},
      "id": "aion-3-0",
      "name": "Aion-3.0",
      "vendor": "AionLabs",
      "released": "2026-07-07",
      "context_window": 131000,
      "price_input": 3,
      "price_output": 6,
      "sources": [
        "https://www.aionlabs.ai/",
        "https://openrouter.ai/aion-labs/aion-3.0",
        "https://writingmate.ai/models/aion-labs/aion-3.0",
        "https://designforonline.com/ai-models/aionlabs-aion-3-0/",
        "https://www.aionlabs.ai/docs/models",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Priciest AionLabs model · 131K ctx · $3/$6",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "aion-labs/aion-3.0",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "writing": {
          "claims": [
            {
              "sentence": "OpenRouter's model listing for Aion-3.0 (checked 2026-09-07) describes it as a multi-model roleplaying and storytelling system built on the GLM family of models.",
              "source_url": "https://openrouter.ai/aion-labs/aion-3.0",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Aion-3.0 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "chat": {
          "claims": [
            {
              "sentence": "OpenRouter's model page for Aion-3.0 (checked 2026-09-07) lists Janitor AI, a character chatbot platform, as its top application by token volume.",
              "source_url": "https://openrouter.ai/aion-labs/aion-3.0",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Janitor AI is a chatbot platform where users create and chat with custom AI characters for interactive roleplay, storytelling, and immersive fiction."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 277,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 260,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "writing"
      ],
      "strengths": [
        "Stronger narrative structure and tension in collaborative storytelling",
        "Nuanced handling of mature and darker themes"
      ],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Aion-3.0-Mini is a multi-model system built on the DeepSeek family, aimed at character-driven fiction and roleplay rather than general reasoning or coding.",
        "Reach for Mini instead of the full Aion-3.0 when you don't need the priciest tier — it targets the same narrative use case on a cheaper DeepSeek-based backbone.",
        "Output tops out at 32K tokens per response, so plan multi-call continuations for longer scenes rather than one giant completion."
      ],
      "task_copy": {},
      "id": "aion-3-0-mini",
      "name": "Aion-3.0-Mini",
      "vendor": "AionLabs",
      "released": "2026-07-07",
      "context_window": 131000,
      "price_input": 0.7,
      "price_output": 1.4,
      "sources": [
        "https://www.aionlabs.ai/",
        "https://openrouter.ai/aion-labs/aion-3.0-mini",
        "https://designforonline.com/ai-models/aionlabs-aion-3-0-mini/",
        "https://www.aionlabs.ai/docs/models",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Cheapest AionLabs model · 131K ctx · $0.7/$1.4",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "aion-labs/aion-3.0-mini",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "writing": {
          "claims": [
            {
              "sentence": "OpenRouter's model listing for Aion-3.0-Mini (checked 2026-09-07) describes it as a multi-model roleplaying and storytelling system built on the DeepSeek family of models.",
              "source_url": "https://openrouter.ai/aion-labs/aion-3.0-mini",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Aion-3.0 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the DeepSeek family of models."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "chat": {
          "claims": [
            {
              "sentence": "OpenRouter's model page for Aion-3.0-Mini (checked 2026-09-07) lists HammerAI, a character-chat app, as its top application by token volume.",
              "source_url": "https://openrouter.ai/aion-labs/aion-3.0-mini",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Chat with AI characters for free"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 321,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 292,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "seed-2-0-code",
      "name": "Seed-2.0-Code",
      "vendor": "ByteDance",
      "released": "2026-02-14",
      "context_window": 256000,
      "sources": [
        "https://seed.bytedance.com/en/seed2",
        "https://howaiworks.ai/models/doubao-seed-code",
        "https://developer.puter.com/ai/bytedance/seed-2.0-code/",
        "https://seed.bytedance.com/en/blog/seed-2-0-official-launch",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "ByteDance model · 256K ctx",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "bytedance-seed/seed-2.0-code",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "ByteDance's official Seed 2.0 launch blog post (checked 2026-09-07) confirms Seed 2.0 includes a dedicated Code model alongside the general-purpose Pro, Lite, and Mini variants.",
              "source_url": "https://seed.bytedance.com/en/blog/seed-2-0-official-launch",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Seed2.0 offers three general-purpose Agent models of different sizes - Pro, Lite, and Mini, along with a dedicated Code model"
            },
            {
              "sentence": "HowAIWorks.ai's model page for Doubao Seed 2.0 Code (checked 2026-09-07) states it has stronger programming and reasoning abilities than ByteDance's preceding Seed coding models.",
              "source_url": "https://howaiworks.ai/models/doubao-seed-code",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Stronger programming and reasoning abilities than the preceding Seed coding models"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "HowAIWorks.ai's model page for Doubao Seed 2.0 Code (checked 2026-09-07) recommends it for agentic feature implementation, described as multi-step work where the model must plan, execute, observe results, and revise the plan.",
              "source_url": "https://howaiworks.ai/models/doubao-seed-code",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "multi-step work where the model must plan, execute, observe results, and revise the plan"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 364,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 360,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "price_input": 0.5,
      "price_output": 3
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Runs on a single Mac with 36GB RAM, so use it for local agentic coding work without needing a cloud API.",
        "Thinking can be toggled per request — enable interleaved thinking between tool calls for harder problems, disable it for quick ones to save latency.",
        "It's text-in/text-out only — route image inputs to a vision-capable model instead."
      ],
      "task_copy": {},
      "id": "laguna-xs-2-1",
      "name": "Laguna XS 2.1",
      "vendor": "Poolside",
      "released": "2026-07-02",
      "context_window": 256000,
      "price_input": 0.06,
      "price_output": 0.12,
      "sources": [
        "https://poolside.ai/blog/introducing-laguna-xs-2-1",
        "https://huggingface.co/poolside/Laguna-XS.2",
        "https://docs.poolside.ai/release-notes/models",
        "https://huggingface.co/poolside/Laguna-XS-2.1-NVFP4",
        "https://benchable.ai/models/poolside/laguna-xs-2.1-20260625",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://huggingface.co/poolside/Laguna-XS-2.1"
      ],
      "best_for_line": "Poolside model · 256K ctx · $0.1/$0.2",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "poolside/laguna-xs-2.1",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Poolside's July 2, 2026 blog post introducing Laguna XS 2.1 (checked 2026-09-07) states it is a 33B total parameter Mixture-of-Experts model with 3B activated parameters per token, up 5.4 points to 63.1% on SWE-bench Multilingual versus the prior XS.2 release.",
              "source_url": "https://poolside.ai/blog/introducing-laguna-xs-2-1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "up 5.4 points to 63.1%"
            },
            {
              "sentence": "Benchmark tracking site Benchable.ai (checked 2026-09-07) reports Laguna XS 2.1 achieving a 100% success rate across all of the benchmarks in its evaluation suite.",
              "source_url": "https://benchable.ai/models/poolside/laguna-xs-2.1-20260625",
              "tier": "measured",
              "date": "2026-09-07",
              "quote": "achieving a 100% success rate across all 7 benchmarks"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Poolside's blog post for Laguna XS 2.1 (checked 2026-09-07) states it is designed for agentic coding and long-horizon work on a local machine.",
              "source_url": "https://poolside.ai/blog/introducing-laguna-xs-2-1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "designed for agentic coding and long-horizon work on a local machine"
            },
            {
              "sentence": "Poolside's Hugging Face model card for the Laguna XS line (checked 2026-09-07) describes it as a lightweight terminal-based coding agent that runs on a Mac with 36GB of RAM.",
              "source_url": "https://huggingface.co/poolside/Laguna-XS-2.1",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "a lightweight terminal-based coding agent"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.05,
          "rank": 95,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 170,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "coding",
        "agentic",
        "research"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Qwen3.8-2.4T-A95B exposes a reasoning_effort parameter — turn it down for routine requests and reserve high effort for genuinely hard coding or research problems, since every call runs a 2.4T-parameter, 95B-active MoE.",
        "Use the preserve_thinking option to retain reasoning context across turns in multi-step agentic sessions instead of re-deriving it from scratch each turn.",
        "Native context is 262K, extensible to roughly 1M, which suits long-horizon agentic and research workflows, but self-hosting needs GB300/B300-class hardware — check hosted API pricing before committing to that scale."
      ],
      "task_copy": {},
      "id": "qwen3-8-2-4t-a95b",
      "name": "Qwen3.8 2.4T A95B",
      "vendor": "Alibaba (Qwen)",
      "released": "2026-08-01",
      "context_window": 983616,
      "price_input": 2,
      "price_output": 6,
      "sources": [
        "https://x.com/Alibaba_Qwen/status/2084100707423289643",
        "https://www.qwencloud.com/models/qwen3.8-2.4t-a95b",
        "https://www.marktechpost.com/2026/07/19/alibaba-previews-qwen3-8-max-a-2-4-trillion-parameter-multimodal-model-days-after-moonshots-kimi-k3-open-weight-launch/",
        "https://developer.nvidia.com/blog/serve-qwen3-8-2-4t-a95b-a-2-4t-parameter-model-with-configurable-reasoning-on-nvidia-gb300-nvl72/",
        "https://huggingface.co/nvidia/Qwen3.8-2.4T-A95B-NVFP4",
        "https://www.siliconrepublic.com/machines/alibaba-unveils-qwen3-8-max-preview"
      ],
      "best_for_line": "Alibaba (Qwen) model · 984K ctx · $2/$6",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "qwen/qwen3.8-2.4t-a95b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://www.qwencloud.com/models/qwen3.8-2.4t-a95b"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Alibaba's own QwenCloud platform page for Qwen3.8-2.4T-A95B lists it as ranked 4th on CodeArena, alongside a GPQA Diamond score of 92.6, as of the page retrieved September 6, 2026.",
              "source_url": "https://www.qwencloud.com/models/qwen3.8-2.4t-a95b",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Ranked 4th on CodeArena."
            },
            {
              "sentence": "Silicon Republic reported on 20 July 2026 that Alibaba unveiled Qwen3.8-Max in preview, describing it as \"second only to\" Anthropic's Fable 5.",
              "source_url": "https://www.siliconrepublic.com/machines/alibaba-unveils-qwen3-8-max-preview",
              "tier": "reported",
              "date": "2026-10-04",
              "quote": "unveiled Qwen3.8-Max in preview - supposedly \"second only to [Anthropic's] Fable 5\""
            }
          ],
          "reconciliation": "Alibaba's July 2026 preview announcement described the model informally as ranking second only to Claude Fable 5 overall; its own later QwenCloud listing gives a narrower, specific claim of 4th place on the CodeArena coding leaderboard, not a general capability ranking.",
          "as_of": "2026-10-04"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.01,
          "rank": 145,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 132,
            "share": 0.03,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "qwen3-8-27b",
      "name": "Qwen3.8 27B",
      "vendor": "Alibaba (Qwen)",
      "released": "2026-08-14",
      "context_window": 262144,
      "price_input": null,
      "price_output": 3,
      "sources": [
        "https://www.alibabacloud.com/blog/alibaba-unveils-qwen3-8-27b-and-releases-weights-of-qwen3-8-flagship-model_603463",
        "https://venturebeat.com/technology/qwen3-8-27b-runs-frontier-class-coding-agents-and-reasoning-locally-no-cloud-api-required",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Alibaba (Qwen) model · 262K ctx",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "qwen/qwen3.8-27b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://www.alibabacloud.com/blog/alibaba-unveils-qwen3-8-27b-and-releases-weights-of-qwen3-8-flagship-model_603463"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Alibaba's Qwen3.8-27B announcement (checked 2026-09-06) states the 27B model delivers excellent capabilities in coding, professional work, research, and long-horizon agentic tasks while matching the performance of the much larger Qwen3.7-plus MoE model.",
              "source_url": "https://www.alibabacloud.com/blog/alibaba-unveils-qwen3-8-27b-and-releases-weights-of-qwen3-8-flagship-model_603463",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Qwen3.8-27B delivers excellent capabilities in handling coding, professional work, research, and long-horizon agentic tasks, matching the performance of Qwen3.7-plus"
            }
          ],
          "reconciliation": "VentureBeat's coverage of Alibaba's own comparison table shows Qwen3.8-27B ahead of Claude Opus 4.6 Max on SWE-bench Pro and LiveCodeBench specifically, but the same table has Opus ahead on Terminal-Bench, GPQA Diamond, and Humanity's Last Exam, so the win is benchmark-specific rather than across the board.",
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Alibaba's Qwen3.8-27B announcement (checked 2026-09-06) states the model is designed to execute complex, multi-step tasks to completion with high reliability, positioning it for agentic workloads.",
              "source_url": "https://www.alibabacloud.com/blog/alibaba-unveils-qwen3-8-27b-and-releases-weights-of-qwen3-8-flagship-model_603463",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "the model is designed to execute complex, multi-step tasks to completion with high reliability."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "research": {
          "claims": [
            {
              "sentence": "Alibaba's Qwen3.8-27B announcement (checked 2026-09-06) lists research among the tasks the model handles with excellent capability while matching Qwen3.7-plus performance.",
              "source_url": "https://www.alibabacloud.com/blog/alibaba-unveils-qwen3-8-27b-and-releases-weights-of-qwen3-8-flagship-model_603463",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Qwen3.8-27B delivers excellent capabilities in handling coding, professional work, research, and long-horizon agentic tasks, matching the performance of Qwen3.7-plus"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.3,
          "rank": 39,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 74,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 30,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 145,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 89,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 40,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 4 rows)",
              "rank": 66,
              "n_models": 140,
              "score": 0.4664,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 74,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 34,
              "n_models": 56,
              "score": 0.475,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 30,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 39,
            "share": 0.51,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 145,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 61,
              "n_models": 83,
              "score": 0.16,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 4 rows)",
              "rank": 69,
              "n_models": 170,
              "score": 0.0543,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 89,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 40,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 20,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "sakana-namazu",
      "name": "Sakana Namazu",
      "vendor": "Sakana AI",
      "released": "2026-08-03",
      "context_window": null,
      "price_input": 0.95,
      "price_output": 4,
      "sources": [
        "https://sakana.ai/namazu-api/",
        "https://aiweekly.co/alerts/sakana-ai-ships-namazu-a-japanese-tuned-kimi-k26-llm",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://sakana.ai/namazu-api"
      ],
      "best_for_line": "Sakana AI model · $0.95/$4",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "sakana/sakana-namazu",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "research": {
          "claims": [
            {
              "sentence": "Sakana AI's Namazu launch page (checked 2026-09-06) describes an automated market-research use case where the model runs an entire research job on its own each Monday morning.",
              "source_url": "https://sakana.ai/namazu-api/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "the first is generating market research reports automatically. a report lands every monday morning, and the model runs the entire job on its own"
            },
            {
              "sentence": "AI Weekly reported (checked 2026-09-06) that Namazu preserves its Kimi K2.6 base model's capabilities on AIME26, MMLU-Pro, and LiveCodeBench v6 while improving sharply on Japanese-specific benchmarks.",
              "source_url": "https://aiweekly.co/alerts/sakana-ai-ships-namazu-a-japanese-tuned-kimi-k26-llm",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "the model preserves kimi's underlying capabilities on aime26, mmlu-pro, and livecodebench v6 while pulling markedly ahead on japanese-specific benchmarks."
            },
            {
              "sentence": "AI Weekly's independent review (checked 2026-09-06) cautioned that Namazu's benchmark gains are a Sakana-run comparison against its own base model, with no head-to-head numbers shown against GPT, Gemini, or Claude.",
              "source_url": "https://aiweekly.co/alerts/sakana-ai-ships-namazu-a-japanese-tuned-kimi-k26-llm",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "the sharpest number here is a sakana-run comparison against its own base model, and the landing page does not show head-to-head benchmarks against gpt, gemini,"
            }
          ],
          "reconciliation": "Sakana's reported benchmark gains (e.g. FairPoliticsQA) are measured against Namazu's own Kimi K2.6 base model, and AI Weekly's 2026-09-06 review confirms no independent head-to-head comparison against GPT, Gemini, or Claude has been published, so the improvement is real but unverified against closed-model peers.",
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Sakana AI's Namazu page (checked 2026-09-06) attributes its long, multi-stage market-research workflow to agentic ability inherited from the Kimi K2.6 base model plus a built-in web-search tool.",
              "source_url": "https://sakana.ai/namazu-api/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "what supports this long, multi-stage workflow is the strong agentic ability inherited from the base model, together with the web search tool that comes"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "writing": {
          "claims": [
            {
              "sentence": "Sakana AI's Namazu launch page (checked 2026-09-06) states the model drafts and writes a market-research report end to end, including web searches and cross-checking, without human involvement.",
              "source_url": "https://sakana.ai/namazu-api/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "from drawing up the research plan through repeated web searches, cross-checking what it finds, and writing the report itself."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "chat": {
          "claims": [
            {
              "sentence": "Sakana AI's Namazu page (checked 2026-09-06) describes the Japanese-specialized model handling customer-support conversations end to end, from answering inquiries to aggregating order data.",
              "source_url": "https://sakana.ai/namazu-api/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "a model specialized for japanese can handle everything end to end, from answering inquiries to aggregating and analyzing order data."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "vision": {
          "claims": [
            {
              "sentence": "Sakana's own Namazu API page (checked 2026-10-16) says the product's demo combines OpenAI-compatible function calling with image recognition, indicating basic vision input support.",
              "source_url": "https://sakana.ai/namazu-api",
              "tier": "lab",
              "date": "2026-10-16",
              "quote": "Behind the demo are OpenAI-compatible function calling and image recognition."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-16"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 444,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 420,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "reasoning",
        "agentic",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Reserve `high` reasoning effort for complex math, coding, and multi-step problems; the default `no_think` mode is faster and cheaper for simple queries.",
        "Use the `low` effort setting as a middle ground when a task needs some reasoning but not the full chain-of-thought cost of `high`.",
        "Hy3's improved reliability on tool calls and output formatting makes it a fit for agentic and coding workflows that need structured output."
      ],
      "task_copy": {},
      "id": "hy3",
      "name": "Hy3",
      "vendor": "Tencent",
      "released": "2026-07-06",
      "context_window": 256000,
      "price_input": 0.15,
      "price_output": 0.59,
      "sources": [
        "https://www.tencent.com/en-us/articles/2202386.html",
        "https://technode.com/2026/07/07/tencent-launches-hunyuan-hy3-integrates-model-across-multiple-products/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://huggingface.co/tencent/Hy3"
      ],
      "best_for_line": "Tencent model · 256K ctx · $0.15/$0.59",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "tencent/hy3",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Tencent's official HY3 launch announcement (checked 2026-09-06) states HY3 made particularly notable progress in productivity tasks including software development and financial modeling compared with the HY3 preview.",
              "source_url": "https://www.tencent.com/en-us/articles/2202386.html",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "hy3 has made particularly notable progress in productivity tasks such as software development, office productivity, financial modeling, front-end design and game production"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Tencent's HY3 launch announcement (checked 2026-09-06) states HY3 improves task planning and tool coordination in agent scenarios, citing WorkBuddy/CodeBuddy, Yuanbao, and Marvis as products using it as an agent.",
              "source_url": "https://www.tencent.com/en-us/articles/2202386.html",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "in agent scenarios, hy3 improves its task planning and tool coordination."
            },
            {
              "sentence": "The same announcement (checked 2026-09-06) specifies HY3 as a 295-billion-total, 21-billion-active-parameter mixture-of-experts model with a 256K context length.",
              "source_url": "https://www.tencent.com/en-us/articles/2202386.html",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "hy3 is a hybrid fast-and-slow-thinking model built on a mixture-of-experts (moe) architecture, with 295 billion total parameters and 21 billion active parameters,"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "frontend": {
          "claims": [
            {
              "sentence": "Tencent's HY3 announcement (checked 2026-09-06) lists front-end design among the productivity tasks where HY3 made particularly notable progress relative to the HY3 preview.",
              "source_url": "https://www.tencent.com/en-us/articles/2202386.html",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "hy3 has made particularly notable progress in productivity tasks such as software development, office productivity, financial modeling, front-end design and game production"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "TechNode reported (checked 2026-09-06) that Tencent prices HY3 API access at RMB 1 ($0.15) per million input tokens and RMB 4 ($0.59) per million output tokens.",
              "source_url": "https://technode.com/2026/07/07/tencent-launches-hunyuan-hy3-integrates-model-across-multiple-products/",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "tencent has priced hy3 at rmb 1 ($0.15) per million input tokens and rmb 4 ($0.59) per million output tokens."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "writing": {
          "claims": [
            {
              "sentence": "Tencent's HY3 announcement (checked 2026-09-06) states HY3 strengthened IMA's knowledge-base Q&A to produce more structured, complete, and usable outputs for long-form writing.",
              "source_url": "https://www.tencent.com/en-us/articles/2202386.html",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "hy3 has also strengthened ima’s knowledge-base q&a to deliver more structured reasoning, broader information coverage, and more complete, structured and usable outputs for long-form writing."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.63,
          "rank": 15,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 61,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 40,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 70,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 60,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 61,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 40,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 19,
            "share": 1.04,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 70,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 60,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "vision",
        "writing"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Thinking Machines' own card calls Inkling \"materially below frontier capabilities\" — save it for lighter conversational and multimodal work, not your hardest reasoning or coding tasks.",
        "It runs locally through the open-source libraries Thinking Machines lists on the model card, or via third-party inference providers — it isn't locked to one API.",
        "Resize images to between 40px and 4096px per dimension before sending them; the model card gives that range for optimal performance.",
        "Keep conversations short — the card documents degraded performance in long multi-turn chats, and warns against medical, legal, or safety-critical use."
      ],
      "task_copy": {},
      "id": "inkling",
      "name": "Inkling",
      "vendor": "Thinking Machines Lab",
      "released": "2026-07-15",
      "context_window": 1048576,
      "price_input": 1,
      "price_output": 4.05,
      "sources": [
        "https://thinkingmachines.ai/model-card/inkling/",
        "https://techcrunch.com/2026/07/15/thinking-machines-amps-up-its-bet-against-one-size-fits-all-ai-with-its-first-open-model-inkling/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://huggingface.co/thinkingmachines/Inkling"
      ],
      "best_for_line": "Thinking Machines Lab model · 1M ctx",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "thinkingmachines/inkling",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "TechCrunch reported (checked 2026-09-06) that on one benchmark, Inkling uses a third as many tokens as Nvidia's Nemotron 3 Ultra to hit the same coding performance, per Thinking Machines.",
              "source_url": "https://techcrunch.com/2026/07/15/thinking-machines-amps-up-its-bet-against-one-size-fits-all-ai-with-its-first-open-model-inkling/",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "inkling uses a third as many tokens as nvidia’s nemotron 3 ultra — its latest generation open-weight model — to hit the same coding performance."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.2,
          "rank": 50,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 77,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 50,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 128,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 83,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 64,
              "n_models": 140,
              "score": 0.4699,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 59,
              "n_models": 125,
              "score": 946,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 127,
              "n_models": 153,
              "score": 0.3229,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 77,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 48,
              "n_models": 56,
              "score": 0.338,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 50,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 68,
            "share": 0.14,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 128,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 59,
              "n_models": 295,
              "score": 0.8826,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 51,
              "n_models": 89,
              "score": 0.403,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 70,
              "n_models": 113,
              "score": 0.3333,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 61,
              "n_models": 70,
              "score": 0.0488,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 83,
              "n_models": 83,
              "score": 0,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 71,
              "n_models": 274,
              "score": 0.8889,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 68,
              "n_models": 170,
              "score": 0.0543,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 65,
              "n_models": 193,
              "score": 0.3653,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 83,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 71,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "vision",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Deploy on Inference Endpoints for up to 160 TPS when you need fast, cheap serving for agentic or chat workloads.",
        "Use its crop/zoom visual-reasoning tools when reading documents or charts where key details are small.",
        "It's also tuned for audio intelligence, so it's a reasonable pick for real-world audio applications, not just text and vision."
      ],
      "task_copy": {},
      "id": "inkling-small",
      "name": "Inkling Small",
      "vendor": "Thinking Machines Lab",
      "released": "2026-08-02",
      "context_window": 1048576,
      "price_input": 0.45,
      "price_output": 1.2,
      "sources": [
        "https://thinkingmachines.ai/news/inkling-small/",
        "https://venturebeat.com/technology/thinking-machines-debuts-inkling-small-open-source-ai-model-nearing-performance-of-predecessor-at-about-1-4-size",
        "https://thinkingmachines.ai/model-card/inkling-small/",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Thinking Machines Lab model · 1M ctx · $0.45/$1.2",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "thinkingmachines/inkling-small",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Thinking Machines' Inkling-Small announcement (checked 2026-09-06) states the model exceeds 80% on SWE-bench Verified, ahead of the 77.6% Inkling's own model card reports for the larger Inkling model on the same benchmark.",
              "source_url": "https://thinkingmachines.ai/news/inkling-small/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "on swebench-verified it exceeds 80%."
            },
            {
              "sentence": "The same announcement (checked 2026-09-06) states Inkling-Small surpassed its larger sibling Inkling on reasoning and agentic coding benchmarks after further agentic-coding reinforcement learning.",
              "source_url": "https://thinkingmachines.ai/news/inkling-small/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "with these improvements, inkling-small surpassed inkling on reasoning and agentic coding benchmarks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Thinking Machines' Inkling-Small announcement (checked 2026-09-06) attributes Inkling-Small's edge over Inkling on agentic benchmarks to continued agentic-coding reinforcement learning built on an on-policy-distilled checkpoint.",
              "source_url": "https://thinkingmachines.ai/news/inkling-small/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "with these improvements, inkling-small surpassed inkling on reasoning and agentic coding benchmarks."
            },
            {
              "sentence": "The announcement (checked 2026-09-06) describes Inkling-Small as achieving comparable performance to the larger Inkling model at about a quarter of its size.",
              "source_url": "https://thinkingmachines.ai/news/inkling-small/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "today, we are releasing inkling-small, an efficient open-weights model that achieves comparable performance to inkling at a quarter of its size."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "vision": {
          "claims": [
            {
              "sentence": "Thinking Machines' Inkling-Small announcement (checked 2026-09-06) states the model features native reasoning over audio and images with a context window of up to 1M tokens.",
              "source_url": "https://thinkingmachines.ai/news/inkling-small/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "like inkling, it features native reasoning over audio and images, variable thinking effort, a context window of up to 1m tokens"
            },
            {
              "sentence": "The same announcement (checked 2026-09-06) states Inkling-Small retains strong performance on visual reasoning, chart and diagram understanding, and visual question answering despite its smaller size.",
              "source_url": "https://thinkingmachines.ai/news/inkling-small/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "it retains strong performance on visual reasoning, chart and diagram understanding, mathematical visual question answering, speech understanding, and longer-form audio reasoning."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.11,
          "rank": 66,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 100,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 48,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 207,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 139,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 70,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 57,
              "n_models": 140,
              "score": 0.4965,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 100,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 48,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 127,
            "share": 0.03,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 207,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 56,
              "n_models": 295,
              "score": 0.8851,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 79,
              "n_models": 89,
              "score": 0.191,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 56,
              "n_models": 113,
              "score": 0.4632,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 53,
              "n_models": 70,
              "score": 0.1707,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best rung: 0.99)",
              "rank": 74,
              "n_models": 83,
              "score": 0.06,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 67,
              "n_models": 274,
              "score": 0.9,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 63,
              "n_models": 170,
              "score": 0.0829,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 62,
              "n_models": 193,
              "score": 0.4014,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 139,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 70,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 72,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "solar-pro-4",
      "name": "Solar Pro 4",
      "vendor": "Upstage",
      "released": "2026-08-10",
      "context_window": 524288,
      "price_input": 0.09,
      "price_output": 0.36,
      "sources": [
        "https://www.upstage.ai/blog/en/solar-pro-4",
        "https://en.sedaily.com/technology/2026/08/14/upstage-unveils-new-ai-model-solar-pro-4",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Upstage model · 524K ctx · $0.3/$1.2",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "upstage/solar-pro4",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "agents": {
          "claims": [
            {
              "sentence": "Seoul Economic Daily reported (checked 2026-09-06) that Solar Pro 4 scored 42 on an Artificial Analysis benchmark, topping the scores of Nvidia's Nemotron 3 Ultra (38) and Google's Gemini 3.5 Flash-Lite (37).",
              "source_url": "https://en.sedaily.com/technology/2026/08/14/upstage-unveils-new-ai-model-solar-pro-4",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "that topped the scores of big-tech models such as nvidia's nemotron 3 ultra (38) and google's gemini 3.5 flash-lite (37)."
            },
            {
              "sentence": "Upstage's own Solar Pro 4 blog post (checked 2026-09-06) reports a Terminal-Bench v2.1 score of 57, describing it as completing multi-step jobs in a live shell rather than just generating commands.",
              "source_url": "https://www.upstage.ai/blog/en/solar-pro-4",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "terminal tasks (terminal-bench v2.1): 57"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "research": {
          "claims": [
            {
              "sentence": "Upstage's Solar Pro 4 blog post (checked 2026-09-06) reports a long-document-reasoning (AA-LCR) score of 71, describing the benchmark as reasoning across roughly 100K tokens of reports and filings to synthesize answers scattered across multiple documents.",
              "source_url": "https://www.upstage.ai/blog/en/solar-pro-4",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "long-document reasoning (aa-lcr): 71"
            },
            {
              "sentence": "Seoul Economic Daily reported (checked 2026-09-06) that Solar Pro 4 scored 71 on the AA-LCR benchmark, more than 2.3 times better than Solar Pro 3's score on the same benchmark.",
              "source_url": "https://en.sedaily.com/technology/2026/08/14/upstage-unveils-new-ai-model-solar-pro-4",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "it scored 71, more than 2.3 times better than the previous model."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "extraction": {
          "claims": [
            {
              "sentence": "Seoul Economic Daily reported (checked 2026-09-06) that Solar Pro 4's AA-LCR benchmark specifically measures the ability to extract information from large volumes of long documents and reason toward answers.",
              "source_url": "https://en.sedaily.com/technology/2026/08/14/upstage-unveils-new-ai-model-solar-pro-4",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "on the long-context understanding benchmark aa-lcr, which measures the ability to extract information from large volumes of long documents and reason toward answers based on"
            },
            {
              "sentence": "Upstage's own blog post (checked 2026-09-06) scores Solar Pro 4 at 71 on this AA-LCR long-document benchmark, sourced from Artificial Analysis.",
              "source_url": "https://www.upstage.ai/blog/en/solar-pro-4",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "long-document reasoning (aa-lcr): 71"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "exec-summaries": {
          "claims": [
            {
              "sentence": "Upstage's Solar Pro 4 blog post (checked 2026-09-06) demonstrates the model turning market data into an Excel workbook, a review report, and a slide deck across three prompts in one session.",
              "source_url": "https://www.upstage.ai/blog/en/solar-pro-4",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "three prompts turned market data into an excel workbook, a review report, and a slide deck."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.26,
          "rank": 43,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 142,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 51,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 236,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 175,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 142,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 51,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 85,
            "share": 0.08,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 236,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 175,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "speed"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Hy-MT2-1.8B is a fast-thinking translation model across 33 languages; use it for high-volume translation jobs rather than general reasoning or coding.",
        "The 1.25-bit AngelSlim quantization shrinks the model to about 440MB and speeds inference roughly 1.5x, making it a good fit for on-device or edge deployment.",
        "It follows terminology, style, and delimiter-preservation instructions given directly in the prompt, so state explicit formatting constraints rather than relying on defaults."
      ],
      "task_copy": {},
      "id": "hy-mt2-1-8b",
      "name": "Hy-MT2-1.8B",
      "vendor": "Tencent",
      "released": "2026-05-21",
      "context_window": 8192,
      "price_input": 0.044,
      "price_output": 0.177,
      "sources": [
        "https://huggingface.co/tencent/Hy-MT2-1.8B",
        "https://github.com/Tencent-Hunyuan/Hy-MT2",
        "https://openrouter.ai/tencent/hy-mt2-1.8b",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Cheapest Tencent model · 8K ctx · $0.044/$0.177",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "bulk": {
          "claims": [
            {
              "sentence": "Tencent's Hy-MT2 GitHub README (checked 2026-09-06) states the 1.8B model surpasses mainstream commercial translation APIs from providers such as Microsoft and Doubao overall, aimed at low-cost, high-volume translation.",
              "source_url": "https://github.com/Tencent-Hunyuan/Hy-MT2",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "the lightweight 1.8b model also surpasses mainstream commercial apis from providers such as microsoft and doubao overall."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 399,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 318,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Hy-MT2-30B-A3B is a 30B mixture-of-experts model with only 3B active parameters, so it stays cheap to run despite outperforming open-source translation models per Tencent's own comparison.",
        "It supports a 256K context window, so use it to translate long documents or multi-turn dialogues in one pass instead of chunking them.",
        "Prefer this 30B-A3B variant over the 1.8B model for the highest translation quality on complex, domain-specific instructions; use the 1.8B for lightweight or on-device jobs instead."
      ],
      "task_copy": {},
      "id": "hy-mt2-30b-a3b",
      "name": "Hy-MT2-30B-A3B",
      "vendor": "Tencent",
      "released": "2026-05-21",
      "context_window": 8192,
      "price_input": 0.074,
      "price_output": 0.295,
      "sources": [
        "https://github.com/Tencent-Hunyuan/Hy-MT2",
        "https://openrouter.ai/tencent/hy-mt2-30b-a3b",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "Tencent model · 8K ctx · $0.074/$0.295",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "tencent/hy-mt2-30b-a3b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "bulk": {
          "claims": [
            {
              "sentence": "Tencent's Hy-MT2 GitHub README (checked 2026-09-06) states the 30B-A3B model outperforms open-source models such as DeepSeek-V4-Pro and Kimi K2.6 in fast-thinking translation mode.",
              "source_url": "https://github.com/Tencent-Hunyuan/Hy-MT2",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "the 7b and 30b-a3b models outperform open-source models such as deepseek-v4-pro and kimi k2.6 in fast-thinking mode"
            },
            {
              "sentence": "OpenRouter's model listing (checked 2026-09-06) describes Hy-MT2-30B-A3B as Tencent's flagship translation model in the Hy-MT2 family, priced at $0.074/$0.295 per million input/output tokens.",
              "source_url": "https://openrouter.ai/tencent/hy-mt2-30b-a3b",
              "tier": "reported",
              "date": "2026-09-06",
              "quote": "hy-mt2-30b-a3b is tencent's flagship translation model in the hy-mt2 family."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 295,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 225,
            "share": 0.01,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "long-context",
        "reasoning"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Reasoning can't be switched off, but effort has three levels — set it to low for simple tasks so you aren't paying for reasoning tokens you don't need.",
        "Save max effort (~75K reasoning tokens) for genuinely complex coding work; Z.ai's own docs recommend low for simpler tasks.",
        "Route non-urgent calls to off-peak hours (all weekend, per the GLM Coding Plan) to pay half the standard points.",
        "Lean on the 1M-token context for whole-repository code review and long-horizon agent tasks rather than chunking files manually."
      ],
      "task_copy": {},
      "id": "glm-5-3",
      "name": "GLM-5.3",
      "vendor": "Z.ai (Zhipu)",
      "released": "2026-08-14",
      "context_window": 1048576,
      "price_input": 1.4,
      "price_output": 4.4,
      "sources": [
        "https://x.com/Zai_org/status/2088132965922476159",
        "https://openrouter.ai/z-ai/glm-5.3",
        "https://datanorth.ai/news/z-ai-releases-glm-5-3",
        "https://docs.z.ai/guides/llm/glm-5.3",
        "https://z.ai/blog/glm-5.3",
        "https://openrouter.ai/api/frontend/v1/rankings/task-spend",
        "https://arena.ai/leaderboard/text"
      ],
      "best_for_line": "Priciest Z.ai (Zhipu) model · 1M ctx · $1.4/$4.4",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "z-ai/glm-5.3",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://docs.z.ai/guides/llm/glm-5.3",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 2.75,
          "rank": 9,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "broad",
      "signals": {
        "coding": {
          "usage_rank": 6,
          "usage_share": 5,
          "arena_rank": 16,
          "expert_default": null,
          "families": 2
        },
        "agents": {
          "usage_rank": 5,
          "usage_share": 5.04,
          "arena_rank": 22,
          "expert_default": null,
          "families": 2
        },
        "bulk": {
          "usage_rank": 7,
          "usage_share": 1.47,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 17,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 20,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 13,
              "n_models": 26,
              "score": 0.426,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 13,
              "n_models": 140,
              "score": 0.5903,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 32,
              "n_models": 125,
              "score": 1317.4,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 20,
              "n_models": 153,
              "score": 0.754,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: max)",
              "rank": 7,
              "n_models": 26,
              "score": 0.6896,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 5,
            "share": 5.94,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 16,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 15,
              "n_models": 66,
              "score": 8163.61,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 18,
              "n_models": 56,
              "score": 0.566,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 5,
            "share": 4.74,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 22,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 13,
            "share": 1.47,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 17,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 33,
              "n_models": 295,
              "score": 0.9091,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 48,
              "n_models": 89,
              "score": 0.41,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 30,
              "n_models": 113,
              "score": 0.6877,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 35,
              "n_models": 70,
              "score": 0.2927,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 34,
              "n_models": 83,
              "score": 0.49,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 63,
              "n_models": 274,
              "score": 0.9111,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 33,
              "n_models": 170,
              "score": 0.1914,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 12,
            "share": 0.58,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 20,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 17,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "qwen3-8-max",
      "name": "Qwen3.8 Max",
      "vendor": "Alibaba (Qwen)",
      "released": "2026-08-03",
      "context_window": 1000000,
      "price_input": 2,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/qwen/qwen3.8-max",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/vision",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "auto_added": "2026-08-28",
      "best_for_line": "Priciest Alibaba (Qwen) model · 1M ctx · $2/$6",
      "availability": {
        "direct_api": null,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.08,
          "rank": 76,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 30,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 16,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 18,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 22,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 3,
          "expert_default": null,
          "families": 1
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 6,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 43,
              "n_models": 140,
              "score": 0.5324,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: xhigh)",
              "rank": 15,
              "n_models": 26,
              "score": 0.5746,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 30,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 14,
              "n_models": 56,
              "score": 0.633,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 16,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 76,
            "share": 0.12,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 18,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 24,
              "n_models": 295,
              "score": 0.9268,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 39,
              "n_models": 89,
              "score": 0.458,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 22,
              "n_models": 113,
              "score": 0.7474,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 25,
              "n_models": 70,
              "score": 0.4634,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 24,
              "n_models": 83,
              "score": 0.58,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 17,
              "n_models": 274,
              "score": 0.9944,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 30,
              "n_models": 170,
              "score": 0.2,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 18,
              "n_models": 42,
              "score": 0.232,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 22,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Vision Arena · Overall (148 models, 1,258,468 votes)",
            "rank": 3,
            "n_models": 148,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/vision"
          }
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 10,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "qwen3-8-flash",
      "name": "Qwen3.8 Flash",
      "vendor": "Alibaba (Qwen)",
      "released": "2026-08-26",
      "context_window": 1000000,
      "price_input": 0.15,
      "price_output": 0.47,
      "sources": [
        "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
        "https://thenewstack.io/qwen38-flash-previews-qwen4/",
        "https://openrouter.ai/qwen/qwen3.8-flash",
        "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/raw/main/README.md"
      ],
      "best_for_line": "Alibaba (Qwen) model · 1M ctx · $0.15/$0.47",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "qwen/qwen3.8-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Qwen's Hugging Face model card for Qwen3.8-Flash-Next, the preview architecture Qwen's own docs say Qwen3.8-Flash is officially built on, reports a SWE-bench Pro score of 62.5, evaluated with the Claude Code harness.",
              "source_url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/raw/main/README.md",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Agentic coding SWE-bench Pro 62.5"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.4,
          "rank": 34,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 32,
            "share": 0.61,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 28,
              "n_models": 42,
              "score": 0.166,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning",
        "research"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Thinking defaults to \"high\" effort out of the box, so it already reasons deeply through math, coding, and planning tasks without you turning anything on.",
        "For quick, low-latency answers, pass reasoning_effort: \"no_think\" via chat_template_kwargs to skip the deep chain-of-thought and get a direct response.",
        "Tencent's own model card flags a tendency to over-verify its own work, spending longer than necessary on simple tasks — watch for this inflating cost.",
        "For production serving, Tencent recommends deploying with vLLM or SGLang rather than a generic inference stack."
      ],
      "task_copy": {},
      "id": "hy4-preview",
      "name": "Hy4 preview",
      "vendor": "Tencent",
      "released": "2026-08-28",
      "context_window": 1048576,
      "price_input": 0.834,
      "price_output": 2.501,
      "sources": [
        "https://www.tencent.com/tencent-releases-and-open-sources-tencent-hy4-preview/",
        "https://technode.com/2026/08/28/tencent-open-sources-hy4-preview-with-770b-parameters-and-a-1m-token-context/",
        "https://openrouter.ai/tencent/hy4-preview",
        "https://huggingface.co/tencent/Hy4-Preview"
      ],
      "best_for_line": "Priciest Tencent model · 1M ctx · $0.834/$2.501",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "tencent/hy4-preview",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "Tencent's own press release (Aug 28, 2026) reports Hy4 preview scoring 2.99 out of 4.00 in an internal blind evaluation of 163 experts across 203 engineering tasks, versus 2.92 for GLM-5.3 and 2.94 for Kimi K3.",
              "source_url": "https://www.tencent.com/tencent-releases-and-open-sources-tencent-hy4-preview/",
              "tier": "lab",
              "date": "2026-08-28",
              "quote": "Hy4 preview scored an average of 2.99 out of 4.00, slightly ahead of GLM-5.3 (2.92/4.00) and Kimi K3 (2.94/4.00)."
            },
            {
              "sentence": "TechNode independently reported (Aug 28, 2026) the same internal evaluation result: Hy4 preview averaged 2.99 out of 4, compared with 2.92 for GLM-5.3 and 2.94 for Kimi K3.",
              "source_url": "https://technode.com/2026/08/28/tencent-open-sources-hy4-preview-with-770b-parameters-and-a-1m-token-context/",
              "tier": "reported",
              "date": "2026-08-28",
              "quote": "Hy4 preview achieved an average score of 2.99 out of 4, compared with 2.92 for GLM 5.3 and 2.94 for Kimi K3."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 4.78,
          "rank": 5,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "preview",
      "adoption": "broad",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 10,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": 5,
          "usage_share": 1.85,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 10,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 10,
            "share": 1.85,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 16,
              "n_models": 83,
              "score": 0.75,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "hy-mt2-7b",
      "name": "Hy-MT2-7B",
      "vendor": "Tencent",
      "released": "2026-05-21",
      "context_window": 8192,
      "price_input": 0.074,
      "price_output": 0.295,
      "sources": [
        "https://huggingface.co/tencent/Hy-MT2-7B",
        "https://github.com/Tencent-Hunyuan/Hy-MT2",
        "https://openrouter.ai/tencent/hy-mt2-7b"
      ],
      "best_for_line": "Tencent model · 8K ctx · $0.074/$0.295",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "tencent/hy-mt2-7b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 440,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 399,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "reasoning"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "id": "granite-4-2-8b",
      "name": "Granite 4.2 8B",
      "vendor": "IBM",
      "released": "2026-08-25",
      "context_window": 128000,
      "sources": [
        "https://research.ibm.com/blog/introducing-granite-4-2",
        "https://huggingface.co/ibm-granite/granite-4.2-8b",
        "https://thenewstack.io/ibm-granite-reasoning-models/",
        "https://openrouter.ai/api/frontend/v1/rankings/models"
      ],
      "best_for_line": "IBM model · 128K ctx",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "ibm-granite/granite-4.2-8b",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "IBM's Granite 4.2 announcement blog states the 8B and 30B models were trained with a dedicated agentic-RL phase aimed at software engineering and terminal-based coding tasks, published 2026-08-25.",
              "source_url": "https://research.ibm.com/blog/introducing-granite-4-2",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "phase focused on enterprise-style tasks, including software engineering, terminal-based coding, and search-driven workflows."
            },
            {
              "sentence": "IBM's own Hugging Face model card for Granite 4.2 8B lists a self-reported SWE-Bench Verified score of 47.67.",
              "source_url": "https://huggingface.co/ibm-granite/granite-4.2-8b",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "47.67"
            },
            {
              "sentence": "The New Stack reported that Qwen 3.8 27B beats the Granite 4.2 models on coding benchmarks and that Granite's coding results are inconsistent.",
              "source_url": "https://thenewstack.io/ibm-granite-reasoning-models/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Qwen 3.8 27B, especially, beats the Granite models across the board, especially when it comes to coding, where the IBM models deliver inconsistent results overall."
            }
          ],
          "reconciliation": "IBM's own SWE-Bench Verified score and agentic-RL training claim real coding ability, but The New Stack's independent review found Qwen 3.8 27B beating Granite 4.2 on coding with inconsistent results, so third-party coding performance is contested rather than clearly strong.",
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "IBM's Granite 4.2 8B model card states the model reasons about which tools to call before invoking them.",
              "source_url": "https://huggingface.co/ibm-granite/granite-4.2-8b",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "The model reasons about which tools to invoke and why, producing more accurate function calls."
            },
            {
              "sentence": "IBM's Granite 4.2 8B model card lists a 512K-token context window built for multi-step agentic workflows.",
              "source_url": "https://huggingface.co/ibm-granite/granite-4.2-8b",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "Supports long documents, multi-turn conversations, and complex agentic workflows."
            },
            {
              "sentence": "The New Stack reported that IBM positions Granite 4.2's main value in high-throughput agentic task performance rather than raw benchmark scores.",
              "source_url": "https://thenewstack.io/ibm-granite-reasoning-models/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "IBM argues that the main argument for Granite is its performance in high-throughput agentic tasks, for example."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "research": {
          "claims": [
            {
              "sentence": "IBM's Granite 4.2 blog describes the models' built-in reasoning as planning before acting, aimed at multi-step analytical work.",
              "source_url": "https://research.ibm.com/blog/introducing-granite-4-2",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "native step-by-step reasoning that helps them plan before they act"
            },
            {
              "sentence": "IBM's Granite 4.2 8B model card reports a self-tested AIME25 score of 86.67.",
              "source_url": "https://huggingface.co/ibm-granite/granite-4.2-8b",
              "tier": "lab",
              "date": "2026-09-07",
              "quote": "86.67"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 196,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 237,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 307,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 278,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 237,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 160,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 307,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 278,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "muse-spark-1-3-contributor",
      "name": "Muse Spark 1.3 Contributor",
      "vendor": "Meta",
      "released": "2026-09-02",
      "context_window": 1048576,
      "price_input": 0.1,
      "price_output": 0.2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "agentic",
        "coding"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-spark-1.3-contributor",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://dev.meta.ai/docs/pricing-rate-limits"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Contributor pricing runs about 90% below Standard ($0.10/$0.20 vs $1.25/$4.25 per M tokens) in exchange for letting Meta train future models on your prompts and completions — keep sensitive data on Standard instead.",
        "Cached input is $0.002/M tokens, far below the $0.10 fresh-input rate — repeat the same long system prompt or context across calls to capture that discount.",
        "Meta positions this tier for prototyping, testing integrations, and scaling experiments where training on your data is acceptable — not for production traffic with sensitive content."
      ],
      "task_copy": {},
      "auto_added": "2026-09-04",
      "best_for_line": "Cheapest Meta model · 1M ctx · $0.1/$0.2",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-spark-1.3-contributor",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter describes Muse Spark 1.3 Contributor as the low-cost tier of Meta's model built for early-stage agentic, multi-agent, and coding workflows.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.3-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "multimodal reasoning model for experimentation, learning, and early-stage agentic, multi-agent, and coding workflows."
            },
            {
              "sentence": "OpenRouter states Muse Spark 1.3 Contributor is designed to track information across extended tasks and ask for clarification when needed.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.3-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It is designed to track information across extended tasks, work through conflicting inputs, and request clarification or confirmation when needed."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "OpenRouter describes Muse Spark 1.3 Contributor as the low-cost tier of Meta's model built for early-stage agentic, multi-agent, and coding workflows.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.3-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "multimodal reasoning model for experimentation, learning, and early-stage agentic, multi-agent, and coding workflows."
            },
            {
              "sentence": "OpenRouter states Muse Spark 1.3 Contributor is designed to track information across extended tasks and ask for clarification when needed.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.3-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It is designed to track information across extended tasks, work through conflicting inputs, and request clarification or confirmation when needed."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.4,
          "rank": 18,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 23,
            "share": 0.77,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "muse-spark-1-3",
      "name": "Muse Spark 1.3",
      "vendor": "Meta",
      "released": "2026-09-02",
      "context_window": 1048576,
      "price_input": 1.25,
      "price_output": 4.25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-spark-1.3",
        "https://www.marktechpost.com/2026/09/03/meta-ai-released-muse-spark-1-3-an-agentic-coding-model-that-uses-20-fewer-tool-calls-and-25-fewer-tokens-than-muse-spark-1-2/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://cursor.com/docs/models/muse-spark-1-3",
        "https://dev.meta.ai/docs/pricing-rate-limits",
        "https://officechai.com/ai/muse-spark-1-3-benchmarks/",
        "https://research.meta.ai/blog/introducing-muse-spark-1-3"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Built for long-running agentic coding — hand it the whole task in one message for long tool-use chains, multi-step debugging, and large-repo work rather than splitting it across turns.",
        "Standard pricing is $1.25/$4.25 per M tokens with a 1,048,576-token context window and $0.15/M cached input — put your stable context first in the prompt to hit the cache rate.",
        "This checkpoint dropped the audio input that 1.1 and 1.2 supported — stay on an older Muse Spark checkpoint if your pipeline still needs audio."
      ],
      "task_copy": {},
      "auto_added": "2026-09-04",
      "best_for_line": "Priciest Meta model · 1M ctx · $1.25/$4.25",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-spark-1.3",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter describes Muse Spark 1.3 as built specifically for long-running coding and agentic workflows.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.3",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows."
            },
            {
              "sentence": "OfficeChai's write-up of Meta's benchmark table (checked 2026-10-04) reports Muse Spark 1.3 at 75.4 on DeepSWE v1.1, ahead of Claude Opus 5 at 74.0.",
              "source_url": "https://officechai.com/ai/muse-spark-1-3-benchmarks/",
              "tier": "reported",
              "date": "2026-10-04",
              "quote": "On DeepSWE v1.1 (long-horizon agentic coding), Muse Spark 1.3 jumps to 75.4 from 1.2's 55.0, edging past Opus 5's 74.0."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-04"
        },
        "agents": {
          "claims": [
            {
              "sentence": "Meta's Muse Spark 1.3 announcement (checked 2026-10-04) says the model asks clarifying questions when prompts are ambiguous and invokes help from the user when stuck.",
              "source_url": "https://research.meta.ai/blog/introducing-muse-spark-1-3",
              "tier": "lab",
              "date": "2026-10-04",
              "quote": "Muse Spark 1.3 asks clarifying questions when prompts are ambiguous, invokes help from the user when stuck, and confirms before taking consequential actions."
            },
            {
              "sentence": "Meta's Muse Spark 1.3 announcement (checked 2026-10-04) says that, in comparisons by Meta engineers against Muse Spark 1.2, it used about 20% fewer tool calls and 25% fewer tokens.",
              "source_url": "https://research.meta.ai/blog/introducing-muse-spark-1-3",
              "tier": "lab",
              "date": "2026-10-04",
              "quote": "In comparisons by Meta engineers, it proved to be significantly faster and more efficient, using ~20% fewer tool calls and ~25% fewer tokens."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-04"
        },
        "bulk": {
          "claims": [
            {
              "sentence": "OfficeChai's write-up of Meta's benchmark table (checked 2026-10-04) reports MRCR scores of 98.5 (256K-512K) and 98.1 (512K-1M) for Muse Spark 1.3, against 91.5 and 73.8 for GPT 5.6 Sol.",
              "source_url": "https://officechai.com/ai/muse-spark-1-3-benchmarks/",
              "tier": "reported",
              "date": "2026-10-04",
              "quote": "On MRCR 256K-512K and MRCR 512K-1M, it scores 98.5 and 98.1, well clear of GPT 5.6 Sol's 91.5 and 73.8"
            },
            {
              "sentence": "OpenRouter's Muse Spark 1.3 page (checked 2026-10-04) lists a 1,048,576-token context window.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.3",
              "tier": "reported",
              "date": "2026-10-04",
              "quote": "Muse Spark 1.3 has a 1,048,576 token context window."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-04"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.05,
          "rank": 96,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "new",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 16,
              "n_models": 26,
              "score": 0.416,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 11,
              "n_models": 140,
              "score": 0.5972,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 15,
              "n_models": 125,
              "score": 1688.35,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror) (best of 2 rows)",
              "rank": 16,
              "n_models": 56,
              "score": 0.586,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 101,
            "share": 0.05,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2 (best of 2 rows)",
              "rank": 23,
              "n_models": 113,
              "score": 0.7439,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2 (best of 2 rows)",
              "rank": 24,
              "n_models": 70,
              "score": 0.4634,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror) (best of 2 rows)",
              "rank": 25,
              "n_models": 83,
              "score": 0.58,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 18,
              "n_models": 274,
              "score": 0.9917,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 19,
              "n_models": 170,
              "score": 0.26,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 10,
              "n_models": 42,
              "score": 0.276,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 11,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "gemini-3-8-flash",
      "name": "Gemini 3.8 Flash",
      "vendor": "Google",
      "released": "2026-09-02",
      "context_window": 1048576,
      "price_input": 0.75,
      "price_output": 3.75,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "reasoning",
        "speed"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/google/gemini-3.8-flash",
        "https://deepmind.google/models/gemini/flash/",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://ai.google.dev/gemini-api/docs/latest-model",
        "https://deepmind.google/models/model-cards/gemini-3-8-flash/"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Stick with the default medium thinking level for most coding and agentic tasks -- Google says it gives the best first-pass accuracy for that class of work.",
        "Drop to low thinking effort for latency-critical jobs like real-time chat, incident response, or fast data analysis, where turnaround matters more than depth.",
        "Save high thinking effort for genuinely hard multi-step reasoning or math -- minimal thinking isn't supported on this model and returns an error if requested.",
        "If token overhead is the binding constraint, use a lower effort level or fall back to 3.7 Flash instead of forcing 3.8 Flash higher."
      ],
      "task_copy": {},
      "auto_added": "2026-09-04",
      "best_for_line": "Google model · 1M ctx · $0.75/$3.75",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "google/gemini-3.8-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://deepmind.google/models/gemini/flash/",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "agents": {
          "claims": [
            {
              "sentence": "Google DeepMind's own product page describes Gemini 3.8 Flash as optimized for high performance across software engineering, long-horizon, and knowledge-work tasks, with reliable agentic execution for coding and real-world issues.",
              "source_url": "https://deepmind.google/models/gemini/flash/",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Versatility across agentic tasks Optimized for high performance across software engineering, long horizon and knowledge work tasks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.83,
          "rank": 24,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 7,
          "expert_default": true,
          "families": 2
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 17,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": 6,
          "usage_share": 1.63,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": 4,
          "usage_share": 4.09,
          "arena_rank": 4,
          "expert_default": null,
          "families": 2
        },
        "research": {
          "usage_rank": 6,
          "usage_share": 3.48,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 2,
          "usage_share": 4.57,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 6,
          "usage_share": 3.12,
          "arena_rank": 8,
          "expert_default": null,
          "families": 2
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 10,
          "usage_share": 2.61,
          "arena_rank": null,
          "expert_default": true,
          "families": 2
        },
        "exec-summaries": {
          "usage_rank": 3,
          "usage_share": 4.35,
          "arena_rank": 3,
          "expert_default": null,
          "families": 2
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: High)",
              "rank": 19,
              "n_models": 26,
              "score": 0.396,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 3 rows)",
              "rank": 23,
              "n_models": 140,
              "score": 0.566,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 37,
              "n_models": 125,
              "score": 1269.72,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 7,
              "n_models": 153,
              "score": 0.8481,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: high)",
              "rank": 2,
              "n_models": 26,
              "score": 0.7383,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 7,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 33,
              "n_models": 66,
              "score": 5093.7933,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 12,
              "n_models": 56,
              "score": 0.643,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 12,
            "share": 0.16,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 17,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 11,
            "share": 1.63,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 4,
            "share": 4.17,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 4,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 5,
              "n_models": 295,
              "score": 0.9539,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 8,
              "n_models": 89,
              "score": 0.697,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 5,
              "n_models": 50,
              "score": 0.4452,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 31,
              "n_models": 113,
              "score": 0.6842,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 47,
              "n_models": 70,
              "score": 0.2195,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 35,
              "n_models": 83,
              "score": 0.48,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 22,
              "n_models": 274,
              "score": 0.9889,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 3 rows)",
              "rank": 34,
              "n_models": 170,
              "score": 0.1829,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 11,
              "n_models": 193,
              "score": 0.8917,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 9,
            "share": 2.46,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 17,
              "n_models": 42,
              "score": 0.234,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 2,
            "share": 4.48,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 3,
            "share": 4.49,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 8,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 10,
            "share": 2.61,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 27,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 3,
            "share": 4.35,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena rank matrix, \"Longer Query\" column (proxy — Arena has no dedicated summarization category)",
            "rank": 3,
            "n_models": 5,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        }
      }
    },
    {
      "id": "gpt-6-astra",
      "name": "GPT-6 Astra",
      "vendor": "OpenAI",
      "released": "2026-09-04",
      "context_window": 1050000,
      "price_input": 10,
      "price_output": 50,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "research"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6-astra",
        "https://platform.openai.com/docs/models/gpt-6-astra.md",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/code/webdev",
        "https://developers.openai.com/api/docs/models/gpt-6-astra",
        "https://openai.com/index/gpt-6-astra/",
        "https://www.foxbusiness.com/technology/openai-unveils-gpt-6-astra-major-advances-ai-capabilities"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "OpenAI positions Astra as state-of-the-art on computer-use/browser-use agentic work, software engineering, and science — reach for it on agentic or coding tasks, not routine chat.",
        "Astra is OpenAI's first model rated 'Critical' for cybersecurity capability, and it rolled out first to a gated trusted-access group — expect staged access rather than full availability at launch."
      ],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "OpenAI model · 1M ctx · $10/$50",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6-astra",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.openai.com/docs/models/gpt-6-astra.md"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenAI's GPT-6 Astra model page (checked 2026-10-03) lists coding among the uses it names for GPT-6 Astra, with complex reasoning, computer use, research and document creation, and a 1,050,000-token context window.",
              "source_url": "https://developers.openai.com/api/docs/models/gpt-6-astra",
              "tier": "lab",
              "date": "2026-10-03",
              "quote": "Use it for complex reasoning, coding, computer use, research, and document creation."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-10-03"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.24,
          "rank": 44,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "new",
      "signals": {
        "coding": {
          "usage_rank": 3,
          "usage_share": 8.88,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": 2,
          "usage_share": 8.38,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": 6,
          "usage_share": 3.75,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 4,
          "usage_share": 4.52,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 6,
          "usage_share": 3.31,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 7,
          "usage_share": 2.97,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 7,
          "usage_share": 3.85,
          "arena_rank": 1,
          "expert_default": null,
          "families": 2
        },
        "exec-summaries": {
          "usage_rank": 2,
          "usage_share": 7.85,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 25,
              "n_models": 140,
              "score": 0.5648,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 1,
              "n_models": 125,
              "score": 2951.32,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 3 rows)",
              "rank": 1,
              "n_models": 153,
              "score": 0.9357,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "DeepSWE (Epoch mirror, mini-swe-agent harness) (best rung: xhigh)",
              "rank": 1,
              "n_models": 26,
              "score": 0.7412,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/deepswe-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 2,
            "share": 9.19,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 2,
              "n_models": 66,
              "score": 15514.7017,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 11,
              "n_models": 56,
              "score": 0.647,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 2,
            "share": 8.12,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 77,
            "share": 0.12,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 6,
            "share": 3.39,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 2,
              "n_models": 295,
              "score": 0.9577,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 2,
              "n_models": 89,
              "score": 0.756,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 1,
              "n_models": 50,
              "score": 0.548,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 2,
              "n_models": 113,
              "score": 0.9368,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2 (best of 6 rows)",
              "rank": 2,
              "n_models": 70,
              "score": 0.976,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 5,
              "n_models": 83,
              "score": 0.99,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 9,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 3,
              "n_models": 170,
              "score": 0.3171,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 2,
              "n_models": 193,
              "score": 0.95,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 3,
            "share": 4.63,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 2,
              "n_models": 42,
              "score": 0.342,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 3,
            "share": 4.16,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 11,
            "share": 0.92,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 7,
            "share": 3.85,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 2,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 2,
            "share": 7.85,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      }
    },
    {
      "id": "gpt-6-astra-pro",
      "name": "GPT-6 Astra Pro",
      "vendor": "OpenAI",
      "released": "2026-09-04",
      "context_window": 1050000,
      "price_input": 10,
      "price_output": 50,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6-astra-pro",
        "https://artificialanalysis.ai/articles/benchmarking-gpt-6-astra",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://thenewstack.io/openai-gpt6-astra-benchmarks/"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "Priciest OpenAI model · 1M ctx · $10/$50",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6-astra-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter states GPT-6 Astra Pro serves the same underlying GPT-6 Astra model at a higher reasoning effort for complex tasks.",
              "source_url": "https://openrouter.ai/openai/gpt-6-astra-pro",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "GPT-6 Astra Pro is the same underlying model as GPT-6 Astra, served with reasoning.mode set to pro for higher-quality responses on complex tasks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "research": {
          "claims": [
            {
              "sentence": "OpenRouter states GPT-6 Astra Pro runs the underlying model at a higher reasoning effort intended for complex analytical tasks.",
              "source_url": "https://openrouter.ai/openai/gpt-6-astra-pro",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "GPT-6 Astra Pro is the same underlying model as GPT-6 Astra, served with reasoning.mode set to pro for higher-quality responses on complex tasks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "The New Stack reported OpenAI's own OSWorld V2-Offline desktop-agent benchmark score for Astra improved over its predecessor GPT-5.6 Sol.",
              "source_url": "https://thenewstack.io/openai-gpt6-astra-benchmarks/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "On the OSWorld V2-Offline benchmark, which tests work across desktop applications, OpenAI says Astra scored 72.6%, up from 65.7% for GPT-5.6 Sol."
            },
            {
              "sentence": "The New Stack reported that Astra cut the average completion time per desktop-agent task nearly in half versus its predecessor.",
              "source_url": "https://thenewstack.io/openai-gpt6-astra-benchmarks/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It also cut the average time per task from about 75 minutes to 40."
            },
            {
              "sentence": "The New Stack reported that Anthropic separately claims a higher OSWorld result for Claude Fable 5.1 than OpenAI's figure for Astra.",
              "source_url": "https://thenewstack.io/openai-gpt6-astra-benchmarks/",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "Anthropic has reported a higher result of 77.9% for Fable 5.1, but"
            }
          ],
          "reconciliation": "OpenAI reports GPT-6 Astra improved its own OSWorld V2-Offline computer-use score to 72.6% over its prior model, but Anthropic separately reported a higher 77.9% for Claude Fable 5.1 on what The New Stack says was a different, non-comparable OSWorld release.",
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.02,
          "rank": 134,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "new",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 3,
          "usage_share": 7.38,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": {
            "rank": 13,
            "share": 0.09,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 210,
            "share": 0.01,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 3,
            "share": 7.38,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "claude-fable-5-1",
      "name": "Claude Fable 5.1",
      "vendor": "Anthropic",
      "released": "2026-09-01",
      "context_window": 1000000,
      "price_input": 10,
      "price_output": 50,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "vision",
        "long-context"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/anthropic/claude-fable-5.1",
        "https://platform.claude.com/docs/en/models/fable-5-1/overview",
        "https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-1",
        "https://www.anthropic.com/claude/fable",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://arena.ai/leaderboard/text",
        "https://arena.ai/leaderboard/agent",
        "https://arena.ai/leaderboard/code/webdev"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Default to effort `high`; save `xhigh` for only your most capability-sensitive workloads — even `low` effort on Fable 5.1 often beats `xhigh` on Claude Opus 4.8.",
        "Prompt-cache reads cost just $0.25/MTok on Fable 5.1 — 2.5% of the input price, versus 10% on other Claude models — so lean harder on caching for long-running or repeated-context agent work.",
        "It declines offensive-cybersecurity and biology/life-sciences requests outright, even benign ones, with stop_reason: refusal — configure a fallback to Claude Opus 4.8 to auto-reroute.",
        "Built for multi-day autonomous runs and parallel subagent delegation — hand it harder tasks than you'd assign prior models and let it scope its own plan."
      ],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "Priciest Anthropic model · 1M ctx · $10/$50",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-fable-5.1",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/models/fable-5-1/overview"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.2,
          "rank": 51,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "new",
      "signals": {
        "coding": {
          "usage_rank": 9,
          "usage_share": 3.52,
          "arena_rank": 34,
          "expert_default": true,
          "families": 3
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 1,
          "expert_default": true,
          "families": 2
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 6,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 9,
          "usage_share": 2.93,
          "arena_rank": null,
          "expert_default": true,
          "families": 2
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 3,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 2,
          "expert_default": true,
          "families": 2
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": 2,
          "expert_default": true,
          "families": 2
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 4,
              "n_models": 26,
              "score": 0.518,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 5 rows)",
              "rank": 3,
              "n_models": 140,
              "score": 0.6308,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 8,
              "n_models": 125,
              "score": 2143.2,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 2,
              "n_models": 153,
              "score": 0.929,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 10,
            "share": 2.73,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Text-Arena \"Coding\" (395 models, 1,680,917 votes)",
            "rank": 34,
            "n_models": 395,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 29,
              "n_models": 66,
              "score": 5421.5567,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror) (best of 2 rows)",
              "rank": 5,
              "n_models": 56,
              "score": 0.686,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 13,
            "share": 0.08,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "Agent Arena · Overall (59 models)",
            "rank": 1,
            "n_models": 59,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/agent"
          }
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 93,
            "share": 0.07,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Creative Writing\" (398 models, 1,231,955 votes)",
            "rank": 6,
            "n_models": 398,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 6,
              "n_models": 89,
              "score": 0.708,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 2,
              "n_models": 50,
              "score": 0.465,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 5,
              "n_models": 113,
              "score": 0.9018,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 7,
              "n_models": 70,
              "score": 0.878,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 1,
              "n_models": 83,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 8,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 7,
              "n_models": 170,
              "score": 0.3114,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 8,
              "n_models": 193,
              "score": 0.9,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 11,
            "share": 1.19,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror) (best of 2 rows)",
              "rank": 8,
              "n_models": 42,
              "score": 0.296,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 12,
            "share": 0.48,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena \"Overall\" (400 models, 7,999,020 votes)",
            "rank": 3,
            "n_models": 400,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 5,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "Text-Arena rank matrix, \"Longer Query\" column (proxy — Arena has no dedicated summarization category)",
            "rank": 2,
            "n_models": 5,
            "as_of": "2026-10-29",
            "url": "https://arena.ai/leaderboard/text"
          }
        }
      }
    },
    {
      "id": "muse-spark-1-2-contributor",
      "name": "Muse Spark 1.2 Contributor",
      "vendor": "Meta",
      "released": "2026-08-21",
      "context_window": 1048576,
      "price_input": 0.1,
      "price_output": 0.2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "cheap-bulk"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-spark-1.2-contributor",
        "https://openrouter.ai/api/frontend/v1/rankings/models",
        "https://gingerlabs.ai/blog/muse-spark-1-2-contributor-api-performance-limits-and-risks"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Muse Spark 1.2 Contributor serves the same weights as Muse Spark 1.2 at roughly 1/12th the price ($0.10/$0.20 per M), but Meta trains on your prompts and outputs — avoid it for proprietary or sensitive code.",
        "Reserve it for low-cost prototyping, large-scale experiments, and agentic coding or tool-heavy workflows where sharing data with Meta for training is acceptable.",
        "It keeps Muse Spark 1.2's 1M-token context window and configurable reasoning effort — turn reasoning effort down for routine tasks to cut cost further on top of the already-discounted rate."
      ],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "Meta model · 1M ctx · $0.1/$0.2",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-spark-1.2-contributor",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "OpenRouter's listing states Muse Spark 1.2 Contributor is a lower-cost Meta reasoning model aimed at developers who want to build cheaply.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.2-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "is a reasoning model from Meta designed for developers who want to start building at an even lower cost."
            },
            {
              "sentence": "OpenRouter reports that in Meta's own testing, Muse Spark 1.2 Contributor performs well on multi-file refactors and whole-repository code generation.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.2-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "it performs well on multi-file refactors, extended debugging sessions, whole-repository generation, and tasks that stretch well past a single prompt."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        },
        "agents": {
          "claims": [
            {
              "sentence": "OpenRouter's listing describes Muse Spark 1.2 Contributor as a Meta reasoning model tailored for complex agentic tasks.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.2-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "It is a reasoning model from Meta, tailored for complex agentic tasks."
            },
            {
              "sentence": "OpenRouter states the model supports structured output, parallel function calling, and configurable reasoning effort for agent use.",
              "source_url": "https://openrouter.ai/meta/muse-spark-1.2-contributor",
              "tier": "reported",
              "date": "2026-09-07",
              "quote": "supports structured output, parallel function calling, and configurable reasoning effort"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.01,
          "rank": 167,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 180,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "deepseek-v4-flash-vision-exp",
      "name": "DeepSeek V4 Flash Vision Exp",
      "vendor": "DeepSeek",
      "released": "2026-08-21",
      "context_window": 1048576,
      "price_input": 0.44,
      "price_output": 1.32,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "vision",
        "agentic"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/deepseek/deepseek-v4-flash-vision-exp",
        "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp/raw/main/README.md",
        "https://api-docs.deepseek.com/news/news260821/",
        "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Reach for this over plain V4-Flash whenever a task involves images, charts, tables, or documents -- DeepSeek's own release notes say it brings multimodal agent performance close to Opus-4.8 while matching V4-Flash on text.",
        "Images are billed at up to 384 tokens each, so batch or downscale screenshots/documents before sending them to keep costs down.",
        "For pure text-only workloads it matches V4-Flash's pricing exactly, so there's no cost penalty to standardizing on the vision-capable variant."
      ],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "DeepSeek model · 1M ctx · $0.22/$0.66",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4-flash-vision-exp",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://api-docs.deepseek.com/news/news260821/",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "DeepSeek's own model card reports DeepSeek-V4-Flash-Vision-Exp scoring 83.9 on Terminal Bench 2.1, slightly above the text-only DeepSeek-V4-Flash-0731's 82.7, while adding multimodal input support.",
              "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp/raw/main/README.md",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "Compared to DeepSeek-V4-Flash-0731, DeepSeek-V4-Flash-Vision-Exp achieves substantial improvements on its multimodal agent capabilities, while maintaining comparable performance on text-only agent tasks."
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.04,
          "rank": 98,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "status": "ga",
      "adoption": "low",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 118,
            "share": 0.04,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "deepseek-v4-pro-0813",
      "name": "DeepSeek V4 Pro 0813",
      "vendor": "DeepSeek",
      "released": "2026-08-12",
      "context_window": 1048576,
      "price_input": 0.58,
      "price_output": 1.73,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/deepseek/deepseek-v4-pro-0813",
        "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813/raw/main/README.md",
        "https://docs.nvidia.com/nim/large-language-models/2.0.12/get-started/advanced/get-started-deepseek-v4-pro-0813.html",
        "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Set reasoning_effort to max for long-horizon agentic runs and complex coding tasks where deliberation quality matters most; drop to low when latency matters more than depth.",
        "For agentic tool-calling, DeepSeek's documented sampling recommendation is top_p=0.95; use top_p=1.0 for non-agentic generation.",
        "It's a 1.6T MoE model with only 49B active params built for sustained multi-step reasoning -- reach for a smaller DeepSeek tier for simple, low-effort requests."
      ],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "DeepSeek model · 1M ctx · $0.58/$1.74",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4-pro-0813",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "DeepSeek's own model card reports DeepSeek-V4-Pro-0813 scoring 87.9 on Terminal Bench 2.1, ahead of Claude Opus-4.8's 85.0 and close to Kimi K3's 88.3, in a table comparing it with DeepSeek-V4-Pro (Preview), GLM-5.2, Kimi K3, and Opus-4.8.",
              "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813/raw/main/README.md",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "DeepSeek-V4-Pro-0813 outperforms DeepSeek-V4-Pro (Preview) on the benchmarks listed below"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": null
      },
      "status": "ga",
      "adoption": "unknown",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 51,
              "n_models": 140,
              "score": 0.5104,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 28,
              "n_models": 125,
              "score": 1403.15,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror)",
              "rank": 27,
              "n_models": 153,
              "score": 0.6622,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 35,
              "n_models": 56,
              "score": 0.473,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 28,
              "n_models": 295,
              "score": 0.9167,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 24,
              "n_models": 89,
              "score": 0.5291,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 38,
              "n_models": 113,
              "score": 0.6456,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 40,
              "n_models": 70,
              "score": 0.2683,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 33,
              "n_models": 83,
              "score": 0.5,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 24,
              "n_models": 274,
              "score": 0.9861,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 35,
              "n_models": 170,
              "score": 0.18,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 44,
              "n_models": 193,
              "score": 0.6125,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 24,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "deepseek-v4-flash-0731",
      "name": "DeepSeek V4 Flash 0731",
      "vendor": "DeepSeek",
      "released": "2026-07-31",
      "context_window": 1310720,
      "price_input": 0.05,
      "price_output": 0.1,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "reasoning"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/deepseek/deepseek-v4-flash-0731",
        "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731/raw/main/README.md"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Use the reasoning_effort parameter (low, high, or max) to control deliberation — max effort needs up to a 384K-token output budget, so size requests accordingly.",
        "DeepSeek's own agentic evals used temperature 1.0 with top_p 0.95 for agentic runs (top_p 1.0 otherwise) — a reasonable starting point when tuning sampling for tool-using tasks.",
        "There's no Jinja chat template; use DeepSeek's own `encoding` folder to convert OpenAI-style messages into model input and to parse outputs.",
        "DeepSeek's serving recipes enable speculative decoding (DSpark) by default (e.g. vLLM's --speculative-config with method dspark) — keep it on for throughput if self-hosting."
      ],
      "task_copy": {},
      "auto_added": "2026-09-06",
      "best_for_line": "Cheapest DeepSeek model · 1M ctx · $0.05/$0.1",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4-flash-0731",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "task_fit_judged": {
        "coding": {
          "claims": [
            {
              "sentence": "DeepSeek's own model card reports DeepSeek-V4-Flash-0731 scoring 82.7 on Terminal Bench 2.1, close to GLM-5.2's 81.0 and Claude Opus-4.8's 85.0, in a table comparing it with DeepSeek-V4-Pro (Preview), GLM-5.2, and Opus-4.8.",
              "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731/raw/main/README.md",
              "tier": "lab",
              "date": "2026-09-06",
              "quote": "DeepSeek-V4-Flash-0731 outperforms DeepSeek-V4-Pro (Preview) on benchmarks listed below despite its far smaller activated parameter count"
            }
          ],
          "reconciliation": null,
          "as_of": "2026-09-07"
        }
      },
      "usage": {
        "openrouter": null
      },
      "status": "ga",
      "adoption": "unknown",
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 56,
              "n_models": 140,
              "score": 0.4988,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 34,
              "n_models": 125,
              "score": 1306.08,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "WeirdML (Epoch mirror) (best of 2 rows)",
              "rank": 32,
              "n_models": 153,
              "score": 0.6296,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/weirdml-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 32,
              "n_models": 295,
              "score": 0.9104,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 61,
              "n_models": 89,
              "score": 0.3363,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 45,
              "n_models": 113,
              "score": 0.5754,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 45,
              "n_models": 70,
              "score": 0.2439,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 27,
              "n_models": 83,
              "score": 0.56,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 48,
              "n_models": 274,
              "score": 0.9444,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 42,
              "n_models": 170,
              "score": 0.1657,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 42,
              "n_models": 193,
              "score": 0.6139,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 25,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      }
    },
    {
      "id": "deepseek-v4-1-flash",
      "name": "DeepSeek V4.1 Flash",
      "vendor": "DeepSeek",
      "released": "2026-09-10",
      "context_window": 1048576,
      "price_input": 0.3,
      "price_output": 1.2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/deepseek/deepseek-v4.1-flash"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 21.12,
          "rank": 1,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-13",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4.1-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": 4,
          "usage_share": 7.08,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": 7,
          "usage_share": 4.04,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": 1,
          "usage_share": 12.19,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": 3,
          "usage_share": 4.5,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 7,
          "usage_share": 3.18,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 5,
          "usage_share": 3.45,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 5,
          "usage_share": 3.36,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 5,
          "usage_share": 5.13,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": 4,
          "usage_share": 4,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 2 rows)",
              "rank": 46,
              "n_models": 140,
              "score": 0.5185,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 52,
              "n_models": 125,
              "score": 1092.35,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 4,
            "share": 6.08,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 44,
              "n_models": 56,
              "score": 0.395,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 7,
            "share": 4.11,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 2,
            "share": 12.19,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 3,
            "share": 4.55,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 47,
              "n_models": 295,
              "score": 0.8984,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 44,
              "n_models": 89,
              "score": 0.429,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 33,
              "n_models": 113,
              "score": 0.6737,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 39,
              "n_models": 70,
              "score": 0.2683,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 29,
              "n_models": 83,
              "score": 0.54,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 26,
              "n_models": 274,
              "score": 0.9833,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 2 rows)",
              "rank": 48,
              "n_models": 170,
              "score": 0.1429,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 25,
              "n_models": 193,
              "score": 0.7292,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 5,
            "share": 3.79,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 23,
              "n_models": 42,
              "score": 0.198,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 7,
            "share": 3.22,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 10,
            "share": 1.03,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 5,
            "share": 5.13,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 18,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 4,
            "share": 4,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "broad",
      "best_for_line": "DeepSeek model · 1M ctx · $0.15/$0.6"
    },
    {
      "id": "grok-4-7",
      "name": "Grok 4.7",
      "vendor": "xAI",
      "released": "2026-09-21",
      "context_window": 500000,
      "price_input": 2,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "coding",
        "agentic",
        "reasoning"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/x-ai/grok-4.7",
        "https://docs.x.ai/developers/grok-4-7"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Set a prompt_cache_key on every request — xAI's docs warn that without it you often pay full input price on a cache-cold server.",
        "Reserve reasoning_effort=xhigh for genuinely hard problems; the default is high, and low/medium cost less for routine tasks.",
        "The Fast variant costs 2x standard rates (1.5x for long-context), so use it only when latency matters more than price.",
        "Long agent loops benefit from context compaction — pair it with function calling for tool-heavy workflows to control token costs."
      ],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.27,
          "rank": 41,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-21",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "x-ai/grok-4.7",
        "aws_bedrock": true,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://docs.x.ai/developers/grok-4-7",
          "https://openrouter.ai/api/v1/models",
          "https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/AmazonBedrock/current/us-east-1/index.json"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Extra High)",
              "rank": 10,
              "n_models": 26,
              "score": 0.463,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 3 rows)",
              "rank": 19,
              "n_models": 140,
              "score": 0.5775,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 12,
            "share": 0.74,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 8,
              "n_models": 66,
              "score": 10536.8317,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 22,
              "n_models": 56,
              "score": 0.546,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 57,
            "share": 0.21,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 23,
              "n_models": 295,
              "score": 0.9274,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 21,
              "n_models": 89,
              "score": 0.56,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 52,
              "n_models": 113,
              "score": 0.5298,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 51,
              "n_models": 70,
              "score": 0.1707,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 47,
              "n_models": 83,
              "score": 0.26,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 30,
              "n_models": 274,
              "score": 0.9806,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 3 rows)",
              "rank": 36,
              "n_models": 170,
              "score": 0.18,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 43,
              "n_models": 193,
              "score": 0.6139,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 20,
              "n_models": 42,
              "score": 0.228,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 12,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Cheapest xAI model · 500K ctx · $1.6/$4.8"
    },
    {
      "id": "gpt-6-luna-pro",
      "name": "GPT-6 Luna Pro",
      "vendor": "OpenAI",
      "released": "2026-09-22",
      "context_window": 1050000,
      "price_input": 0.1,
      "price_output": 0.5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6-luna-pro"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.26,
          "rank": 42,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6-luna-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 44,
            "share": 0.39,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Cheapest OpenAI model · 1M ctx · $0.1/$0.5"
    },
    {
      "id": "gpt-6-luna",
      "name": "GPT-6 Luna",
      "vendor": "OpenAI",
      "released": "2026-09-22",
      "context_window": 1050000,
      "price_input": 0.1,
      "price_output": 0.5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6-luna"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 5.21,
          "rank": 4,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6-luna",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": 2,
          "usage_share": 5.98,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": 6,
          "usage_share": 3.29,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 34,
              "n_models": 140,
              "score": 0.5463,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 22,
              "n_models": 125,
              "score": 1576.92,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 40,
              "n_models": 56,
              "score": 0.443,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 4,
            "share": 5.98,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 42,
              "n_models": 295,
              "score": 0.9047,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 46,
              "n_models": 89,
              "score": 0.414,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 19,
              "n_models": 113,
              "score": 0.7895,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 19,
              "n_models": 70,
              "score": 0.561,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 20,
              "n_models": 83,
              "score": 0.64,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 21,
              "n_models": 274,
              "score": 0.9889,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 32,
              "n_models": 170,
              "score": 0.1943,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 50,
              "n_models": 193,
              "score": 0.5931,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 19,
              "n_models": 42,
              "score": 0.23,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 23,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 6,
            "share": 3.29,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "broad",
      "best_for_line": "OpenAI model · 1M ctx · $0.1/$0.5"
    },
    {
      "id": "gpt-6-sol-pro",
      "name": "GPT-6 Sol Pro",
      "vendor": "OpenAI",
      "released": "2026-09-22",
      "context_window": 1050000,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6-sol-pro"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 211,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6-sol-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 289,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "OpenAI model · 1M ctx · $2/$10"
    },
    {
      "id": "gpt-6-sol",
      "name": "GPT-6 Sol",
      "vendor": "OpenAI",
      "released": "2026-09-22",
      "context_window": 1050000,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6-sol"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.15,
          "rank": 57,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6-sol",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 20,
              "n_models": 140,
              "score": 0.5764,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 4,
              "n_models": 125,
              "score": 2462,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 3,
              "n_models": 66,
              "score": 14427.8483,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 25,
              "n_models": 56,
              "score": 0.543,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 10,
            "share": 0.33,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 88,
            "share": 0.08,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 10,
              "n_models": 295,
              "score": 0.9432,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 15,
              "n_models": 89,
              "score": 0.607,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 6,
              "n_models": 113,
              "score": 0.8982,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 6,
              "n_models": 70,
              "score": 0.9,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 14,
              "n_models": 83,
              "score": 0.83,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 5,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 8,
              "n_models": 170,
              "score": 0.3086,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 9,
              "n_models": 193,
              "score": 0.8958,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 11,
              "n_models": 42,
              "score": 0.264,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 6,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "OpenAI model · 1M ctx · $2/$10"
    },
    {
      "id": "claude-opus-5-5",
      "name": "Claude Opus 5.5",
      "vendor": "Anthropic",
      "released": "2026-09-22",
      "context_window": 1000000,
      "price_input": 4,
      "price_output": 20,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "coding",
        "research"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/anthropic/claude-opus-5.5",
        "https://platform.claude.com/docs/en/models/overview"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Anthropic positions Opus 5.5 for long-running agentic coding and knowledge work, not quick low-stakes queries.",
        "Prompt cache reads cost just 5% of the base input price on Opus 5.5 (vs 10% on Haiku 5.5), so caching tool/system prompts pays off fast on long agent loops.",
        "If Opus 5.5 at higher effort still falls short on your evals, Anthropic's own guidance is to step up to Claude Fable 5.1 rather than raising effort further.",
        "Batch API requests get 50% off list price, worth using for large, non-interactive knowledge-work jobs."
      ],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 4.2,
          "rank": 6,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-22",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-opus-5.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/models/overview"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": 1,
          "usage_share": 19.64,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": 1,
          "usage_share": 22.05,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": 8,
          "usage_share": 1.46,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": 1,
          "usage_share": 10.98,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 1,
          "usage_share": 21.25,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 1,
          "usage_share": 10.89,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 1,
          "usage_share": 24.07,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 1,
          "usage_share": 17.04,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": 1,
          "usage_share": 12.5,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 1,
              "n_models": 26,
              "score": 0.578,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 5 rows)",
              "rank": 1,
              "n_models": 140,
              "score": 0.669,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 7,
              "n_models": 125,
              "score": 2147.38,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 1,
            "share": 16.98,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 11,
              "n_models": 66,
              "score": 9235.2483,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror) (best of 2 rows)",
              "rank": 3,
              "n_models": 56,
              "score": 0.735,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 1,
            "share": 20.7,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 14,
            "share": 1.46,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 1,
            "share": 10.33,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 39,
              "n_models": 295,
              "score": 0.9059,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 5,
              "n_models": 89,
              "score": 0.722,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 3,
              "n_models": 113,
              "score": 0.9123,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 3,
              "n_models": 70,
              "score": 0.95,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 2,
              "n_models": 83,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 6,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 4,
              "n_models": 170,
              "score": 0.3171,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 4,
              "n_models": 193,
              "score": 0.9333,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 1,
            "share": 15.42,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 6,
              "n_models": 42,
              "score": 0.306,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 1,
            "share": 11.14,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 1,
            "share": 11.57,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 1,
            "share": 17.04,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 1,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 1,
            "share": 12.5,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "broad",
      "best_for_line": "Anthropic model · 1M ctx · $4/$20"
    },
    {
      "id": "qwen3-8-omni-flash",
      "name": "Qwen3.8 Omni Flash",
      "vendor": "Alibaba (Qwen)",
      "released": "2026-09-21",
      "context_window": 1000000,
      "price_input": 0.15,
      "price_output": 0.47,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/qwen/qwen3.8-omni-flash"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.02,
          "rank": 126,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "qwen/qwen3.8-omni-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 87,
            "share": 0.08,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Alibaba (Qwen) model · 1M ctx · $0.15/$0.47"
    },
    {
      "id": "command-a",
      "name": "Command A+",
      "vendor": "Cohere",
      "released": "2026-09-22",
      "context_window": 192000,
      "price_input": 0.3,
      "price_output": 1.5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "agentic",
        "vision",
        "speed"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/cohere/command-a-plus",
        "https://docs.cohere.com/v1/docs/command-a-plus"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Command A+ is built for multimodal agentic tasks, so reach for it when a workflow needs the model to take multi-step actions over text and image inputs, not just answer a question.",
        "It covers 48 languages, including every official EU language, making it a solid default for multilingual agent deployments without swapping models per locale.",
        "Cohere designed it to run on one B200 or two H100s at W4A4, so it needs fewer GPUs per instance than larger agentic models — worth factoring in if you're self-hosting."
      ],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": null
      },
      "auto_added": "2026-09-23",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "cohere/command-a",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Cohere model · 192K ctx · $0.3/$1.5"
    },
    {
      "id": "qwen3-8-max-prime",
      "name": "Qwen3.8 Max Prime",
      "vendor": "Alibaba (Qwen)",
      "released": "2026-09-23",
      "context_window": 1000000,
      "price_input": 4,
      "price_output": 12,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/qwen/qwen3.8-max-prime"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.02,
          "rank": 127,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-23",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "qwen/qwen3.8-max-prime",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 157,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest Alibaba (Qwen) model · 1M ctx · $4/$12"
    },
    {
      "id": "claude-sonnet-5-5",
      "name": "Claude Sonnet 5.5",
      "vendor": "Anthropic",
      "released": "2026-09-28",
      "context_window": 1000000,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/anthropic/claude-sonnet-5.5"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.32,
          "rank": 19,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-28",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-sonnet-5.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": 7,
          "usage_share": 4.7,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": 4,
          "usage_share": 5.14,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": 10,
          "usage_share": 1.05,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": 2,
          "usage_share": 7.68,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 3,
          "usage_share": 5.07,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 3,
          "usage_share": 4.13,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": 3,
          "usage_share": 5,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 6,
          "usage_share": 4.34,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": 5,
          "usage_share": 3.9,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 2,
              "n_models": 26,
              "score": 0.555,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 5 rows)",
              "rank": 6,
              "n_models": 140,
              "score": 0.61,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 14,
              "n_models": 125,
              "score": 1819.12,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 8,
            "share": 4.29,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 6,
              "n_models": 66,
              "score": 11165.2783,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror) (best of 2 rows)",
              "rank": 2,
              "n_models": 56,
              "score": 0.755,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 4,
            "share": 4.87,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 18,
            "share": 1.05,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 2,
            "share": 7.28,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 3,
              "n_models": 295,
              "score": 0.9558,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 37,
              "n_models": 89,
              "score": 0.465,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 4,
              "n_models": 50,
              "score": 0.4612,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 8,
              "n_models": 113,
              "score": 0.8877,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 9,
              "n_models": 70,
              "score": 0.8049,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 3,
              "n_models": 83,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 4,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 6,
              "n_models": 170,
              "score": 0.3143,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 18,
              "n_models": 193,
              "score": 0.8389,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 4,
            "share": 4.25,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 3,
              "n_models": 42,
              "score": 0.314,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 4,
            "share": 3.59,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": {
            "rank": 5,
            "share": 2.54,
            "tags": [
              "conversational_reply",
              "customer_support",
              "roleplay_fiction"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 6,
            "share": 4.34,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 3,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 5,
            "share": 3.9,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "best_for_line": "Anthropic model · 1M ctx · $2/$10"
    },
    {
      "id": "gpt-6-1-sol-pro",
      "name": "GPT-6.1 Sol Pro",
      "vendor": "OpenAI",
      "released": "2026-09-29",
      "context_window": 1050000,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6.1-sol-pro"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.03,
          "rank": 108,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-29",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6.1-sol-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 152,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "OpenAI model · 1M ctx · $2/$10"
    },
    {
      "id": "gpt-6-1-sol",
      "name": "GPT-6.1 Sol",
      "vendor": "OpenAI",
      "released": "2026-09-29",
      "context_window": 1050000,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6.1-sol"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.64,
          "rank": 14,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-09-29",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6.1-sol",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": 10,
          "usage_share": 3.24,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "agents": {
          "usage_rank": 8,
          "usage_share": 3.49,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": 9,
          "usage_share": 2.63,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "research": {
          "usage_rank": 8,
          "usage_share": 3.04,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "extraction": {
          "usage_rank": 7,
          "usage_share": 3.17,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": 9,
          "usage_share": 2.75,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "exec-summaries": {
          "usage_rank": 8,
          "usage_share": 2.97,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 5 rows)",
              "rank": 29,
              "n_models": 140,
              "score": 0.5579,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 2,
              "n_models": 125,
              "score": 2743.75,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 9,
            "share": 3.92,
            "tags": [
              "code:general_impl",
              "code:debugging",
              "code:file_read_write",
              "code:review_security",
              "code:repo_scan",
              "code:sql_database",
              "code:devops_config"
            ],
            "n_models": 15,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 15,
              "n_models": 56,
              "score": 0.6,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 9,
            "share": 3.35,
            "tags": [
              "agent:workflow_execution",
              "agent:multi_step_planning",
              "agent:web_search",
              "agent:tool_dispatch",
              "agent:memory_extraction"
            ],
            "n_models": 14,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 20,
            "share": 1.04,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": {
            "rank": 9,
            "share": 2.38,
            "tags": [
              "content_writing",
              "translation"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 4,
              "n_models": 295,
              "score": 0.9539,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 3,
              "n_models": 89,
              "score": 0.739,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 1,
              "n_models": 113,
              "score": 0.9368,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 1,
              "n_models": 70,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 6,
              "n_models": 83,
              "score": 0.99,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 3,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 5,
              "n_models": 170,
              "score": 0.3171,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 3,
              "n_models": 193,
              "score": 0.9417,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 6,
            "share": 3.5,
            "tags": [
              "research_report",
              "qa_knowledge",
              "math"
            ],
            "n_models": 13,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 1,
              "n_models": 42,
              "score": 0.35,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": {
            "rank": 5,
            "share": 3.57,
            "tags": [
              "data:extraction",
              "data:transformation",
              "classification_tagging"
            ],
            "n_models": 12,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": {
            "rank": 9,
            "share": 2.75,
            "tags": [
              "code:frontend_ui"
            ],
            "n_models": 10,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 4,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": {
            "rank": 8,
            "share": 2.97,
            "tags": [
              "summarization"
            ],
            "n_models": 9,
            "as_of": "2026-10-29",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "best_for_line": "OpenAI model · 1M ctx · $2/$10"
    },
    {
      "id": "mistral-large-4",
      "name": "Mistral Large 4",
      "vendor": "Mistral AI",
      "released": "2026-10-06",
      "context_window": 524288,
      "price_input": 0.68,
      "price_output": 2.09,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "reasoning",
        "agentic",
        "coding",
        "vision",
        "research"
      ],
      "strengths": [
        "Software engineering, repository understanding, and complex terminal/agentic workflows.",
        "Document, chart, and image understanding with visual grounding on technical and satellite imagery.",
        "Scientific coding (simulations, data analysis) and applied math."
      ],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/mistralai/mistral-large-4-0",
        "https://mistral.ai/news/mistral-large-4",
        "https://docs.mistral.ai/getting-started/models/models_overview/"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "For cybersecurity work (vulnerability analysis, malware analysis, detection-rule writing), Mistral positions Large 4 for private-cloud or on-premise deployment rather than the public API.",
        "It reads documents, charts, and images directly, including visual grounding on engineering drawings and satellite imagery, so skip separate OCR preprocessing for those inputs.",
        "It's a hybrid instruct-and-reasoning model meant to unify plain instruction-following, reasoning, and agentic tool use in one model rather than needing a separate reasoning-tier model.",
        "Self-hosted weights aren't out yet (Mistral said they're due by end of October 2026), so for now plan on the hosted Mistral Studio preview API rather than local deployment."
      ],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": null
      },
      "auto_added": "2026-10-06",
      "availability": {
        "direct_api": true,
        "openrouter": false,
        "openrouter_id": null,
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://mistral.ai/news/mistral-large-4",
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": 100,
          "basis": [
            "best_for:vision"
          ]
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest Mistral AI model · 524K ctx · $0.68/$2.09"
    },
    {
      "id": "claude-haiku-5-5",
      "name": "Claude Haiku 5.5",
      "vendor": "Anthropic",
      "released": "2026-10-07",
      "context_window": 1000000,
      "price_input": 0.1,
      "price_output": 0.5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [
        "cheap-bulk",
        "speed"
      ],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/anthropic/claude-haiku-5.5",
        "https://platform.claude.com/docs/en/models/overview"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [
        "Anthropic's own models page puts Haiku 5.5 to work on high-volume, latency-sensitive tasks such as classification, extraction, and routing, not deep reasoning.",
        "Batch API requests are 50% off list price and prompt cache reads cost just 10% of the base input price, so cache repeated system prompts on bulk Haiku jobs.",
        "Haiku 5.5 defaults to medium thinking effort; most high-volume classification and routing work does not need to raise it."
      ],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.56,
          "rank": 16,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-07",
      "availability": {
        "direct_api": true,
        "openrouter": true,
        "openrouter_id": "anthropic/claude-haiku-5.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models",
          "https://platform.claude.com/docs/en/models/overview"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": 4,
          "usage_share": 3.06,
          "arena_rank": null,
          "expert_default": null,
          "families": 1
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 6,
              "n_models": 26,
              "score": 0.484,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 5 rows)",
              "rank": 33,
              "n_models": 140,
              "score": 0.5498,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 21,
              "n_models": 125,
              "score": 1587,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 44,
              "n_models": 66,
              "score": 3602.4633,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 20,
              "n_models": 56,
              "score": 0.561,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 9,
            "share": 3.06,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 49,
              "n_models": 295,
              "score": 0.8965,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 73,
              "n_models": 89,
              "score": 0.238,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "Humanity's Last Exam (Epoch mirror)",
              "rank": 13,
              "n_models": 50,
              "score": 0.3012,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/hle-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 21,
              "n_models": 113,
              "score": 0.7509,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 23,
              "n_models": 70,
              "score": 0.4634,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 10,
              "n_models": 83,
              "score": 0.89,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025 (best of 2 rows)",
              "rank": 20,
              "n_models": 274,
              "score": 0.9889,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 5 rows)",
              "rank": 25,
              "n_models": 170,
              "score": 0.2257,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 45,
              "n_models": 193,
              "score": 0.6125,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "best_for_line": "Cheapest Anthropic model · 1M ctx · $0.1/$0.5"
    },
    {
      "id": "gemini-4-argon",
      "name": "Gemini 4 Argon",
      "vendor": "Google",
      "released": "2026-10-13",
      "context_window": 1048576,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/google/gemini-4-argon"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.49,
          "rank": 32,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-13",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "google/gemini-4-argon",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: High)",
              "rank": 3,
              "n_models": 26,
              "score": 0.529,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 4,
              "n_models": 140,
              "score": 0.6181,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror) (best of 2 rows)",
              "rank": 9,
              "n_models": 125,
              "score": 2100.1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 4,
              "n_models": 66,
              "score": 13718.1583,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 1,
              "n_models": 56,
              "score": 0.824,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 42,
            "share": 0.44,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 1,
              "n_models": 295,
              "score": 0.964,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 1,
              "n_models": 89,
              "score": 0.76,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 11,
              "n_models": 113,
              "score": 0.8596,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 10,
              "n_models": 70,
              "score": 0.7805,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 7,
              "n_models": 83,
              "score": 0.99,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 2,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 17,
              "n_models": 170,
              "score": 0.2714,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 6,
              "n_models": 193,
              "score": 0.9125,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GDP.pdf (Epoch mirror)",
              "rank": 9,
              "n_models": 42,
              "score": 0.284,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gdp-pdf-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": {
            "board": "WebDev Arena (Epoch mirror) (mirrors arena.ai's blind human votes)",
            "rank": 8,
            "n_models": 120,
            "as_of": "2026-10-29",
            "url": "https://epoch.ai/benchmarks/webdev-arena-external",
            "licence": "signal-only",
            "score": null
          }
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest Google model · 1M ctx · $2/$10"
    },
    {
      "id": "deepseek-v4-1-flash-1017",
      "name": "DeepSeek V4.1 Flash 1017",
      "vendor": "DeepSeek",
      "released": "2026-10-17",
      "context_window": 1048576,
      "price_input": 0.15,
      "price_output": 0.6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/deepseek/deepseek-v4.1-flash-1017"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": null
      },
      "auto_added": "2026-10-17",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "deepseek/deepseek-v4.1-flash-1017",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "CursorBench (best rung: Max)",
              "rank": 8,
              "n_models": 26,
              "score": 0.477,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/cursorbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 28,
              "n_models": 140,
              "score": 0.559,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 18,
              "n_models": 125,
              "score": 1622,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 4,
              "n_models": 56,
              "score": 0.698,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 20,
              "n_models": 295,
              "score": 0.9318,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 45,
              "n_models": 89,
              "score": 0.425,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 27,
              "n_models": 113,
              "score": 0.7123,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 22,
              "n_models": 70,
              "score": 0.4634,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 15,
              "n_models": 274,
              "score": 0.9944,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 22,
              "n_models": 170,
              "score": 0.24,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            },
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 20,
              "n_models": 193,
              "score": 0.7806,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "DeepSeek model · 1M ctx · $0.15/$0.6"
    },
    {
      "id": "muse-spark-open",
      "name": "Muse Spark Open",
      "vendor": "Meta",
      "released": "2026-10-15",
      "context_window": 1048576,
      "price_input": 1.25,
      "price_output": 4.25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-spark-open"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.03,
          "rank": 105,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-spark-open",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 135,
            "share": 0.03,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "arc-prize",
              "benchmark": "ARC-AGI-2 (semi-private evaluation)",
              "rank": 52,
              "n_models": 193,
              "score": 0.575,
              "as_of": "2026-10-29",
              "url": "https://arcprize.org/leaderboard",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Meta model · 1M ctx · $1.25/$4.25"
    },
    {
      "id": "nemotron-3-5-super-vl",
      "name": "Nemotron 3.5 Super VL",
      "vendor": "NVIDIA",
      "released": "2026-10-15",
      "context_window": 1048576,
      "price_input": 0.15,
      "price_output": 0.5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/nvidia/nemotron-3.5-super-vl"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 1.03,
          "rank": 22,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "nvidia/nemotron-3.5-super-vl",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 28,
            "share": 0.69,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "best_for_line": "Priciest NVIDIA model · 1M ctx · $0.15/$0.5"
    },
    {
      "id": "muse-wonder",
      "name": "Muse Wonder",
      "vendor": "Meta",
      "released": "2026-10-12",
      "context_window": 1048576,
      "price_input": 4,
      "price_output": 20,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/meta/muse-wonder"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.05,
          "rank": 93,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "meta/muse-wonder",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 154,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest Meta model · 1M ctx · $4/$20"
    },
    {
      "id": "ling-3-1-flash",
      "name": "Ling 3.1 Flash",
      "vendor": "InclusionAI (Ant Group)",
      "released": "2026-10-02",
      "context_window": 262144,
      "price_input": 0.08,
      "price_output": 0.22,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/inclusionai/ling-3.1-flash"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.01,
          "rank": 138,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "inclusionai/ling-3.1-flash",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 112,
            "share": 0.04,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest InclusionAI (Ant Group) model · 262K ctx · $0.08/$0.22"
    },
    {
      "id": "glm-5-3-prime",
      "name": "GLM 5.3 Prime",
      "vendor": "Z.ai (Zhipu)",
      "released": "2026-09-23",
      "context_window": 1000000,
      "price_input": 2.8,
      "price_output": 8.8,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/z-ai/glm-5.3-prime"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.02,
          "rank": 119,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "z-ai/glm-5.3-prime",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 173,
            "share": 0.02,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest Z.ai (Zhipu) model · 1M ctx · $2.8/$8.8"
    },
    {
      "id": "aion-3-5-mini",
      "name": "Aion 3.5 Mini",
      "vendor": "AionLabs",
      "released": "2026-09-23",
      "context_window": 262144,
      "price_input": 0.7,
      "price_output": 1.4,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/aion-labs/aion-3.5-mini"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 285,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "aion-labs/aion-3.5-mini",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 249,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "AionLabs model · 262K ctx · $0.7/$1.4"
    },
    {
      "id": "aion-3-5",
      "name": "Aion 3.5",
      "vendor": "AionLabs",
      "released": "2026-09-23",
      "context_window": 262144,
      "price_input": 3,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/aion-labs/aion-3.5"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 256,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "aion-labs/aion-3.5",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 248,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest AionLabs model · 262K ctx · $3/$6"
    },
    {
      "id": "solar-mini-4",
      "name": "Solar Mini 4",
      "vendor": "Upstage",
      "released": "2026-09-23",
      "context_window": 524288,
      "price_input": 0.05,
      "price_output": 0.2,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/upstage/solar-mini4"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.73,
          "rank": 26,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "upstage/solar-mini4",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 47,
            "share": 0.3,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "best_for_line": "Cheapest Upstage model · 524K ctx · $0.05/$0.2"
    },
    {
      "id": "glm-5-3-flashx",
      "name": "GLM 5.3 FlashX",
      "vendor": "Z.ai (Zhipu)",
      "released": "2026-09-18",
      "context_window": 1048576,
      "price_input": 0.37,
      "price_output": 1.25,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/z-ai/glm-5.3-flashx"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.07,
          "rank": 78,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "z-ai/glm-5.3-flashx",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 102,
            "share": 0.05,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Cheapest Z.ai (Zhipu) model · 1M ctx · $0.37/$1.25"
    },
    {
      "id": "fugu-ultra-v2",
      "name": "Fugu Ultra v2",
      "vendor": "Sakana AI",
      "released": "2026-09-11",
      "context_window": 1000000,
      "price_input": 5,
      "price_output": 30,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/sakana/fugu-ultra-v2"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 386,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-21",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "sakana/fugu-ultra-v2",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 368,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Priciest Sakana AI model · 1M ctx · $5/$30"
    },
    {
      "id": "fugu-max",
      "name": "Fugu Max",
      "vendor": "Sakana AI",
      "released": "2026-09-11",
      "context_window": 1000000,
      "price_input": 2,
      "price_output": 6,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/sakana/fugu-max"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 355,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "sakana/fugu-max",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 356,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "Sakana AI model · 1M ctx · $2/$6"
    },
    {
      "id": "ling-3-0-flash-vl",
      "name": "Ling 3.0 Flash VL",
      "vendor": "InclusionAI (Ant Group)",
      "released": "2026-09-10",
      "context_window": 131072,
      "price_input": 0.06,
      "price_output": 0.18,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/inclusionai/ling-3.0-flash-vl"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 313,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-22",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "inclusionai/ling-3.0-flash-vl",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": true,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 266,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "InclusionAI (Ant Group) model · 131K ctx · $0.06/$0.18"
    },
    {
      "id": "kimi-k3-1-pro",
      "name": "Kimi K3.1 Pro",
      "vendor": "Moonshot AI",
      "released": "2026-10-23",
      "context_window": 1048576,
      "price_input": 2,
      "price_output": 10,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/moonshotai/kimi-k3.1-pro"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.67,
          "rank": 27,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-23",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "moonshotai/kimi-k3.1-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror)",
              "rank": 8,
              "n_models": 140,
              "score": 0.6076,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 13,
              "n_models": 125,
              "score": 1932.75,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 49,
            "share": 0.29,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "ProofBench (Epoch mirror)",
              "rank": 9,
              "n_models": 83,
              "score": 0.91,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/proofbench-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror)",
              "rank": 21,
              "n_models": 170,
              "score": 0.2457,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "moderate",
      "best_for_line": "Moonshot AI model · 1M ctx · $2/$10"
    },
    {
      "id": "gpt-6-1-terra",
      "name": "GPT-6.1 Terra",
      "vendor": "OpenAI",
      "released": "2026-10-26",
      "context_window": 1050000,
      "price_input": 1,
      "price_output": 5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6.1-terra"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0.09,
          "rank": 71,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-26",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6.1-terra",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "SciCode (Epoch mirror) (best of 6 rows)",
              "rank": 30,
              "n_models": 140,
              "score": 0.5509,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/scicode-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "ALE-Bench (Epoch mirror)",
              "rank": 3,
              "n_models": 125,
              "score": 2608.82,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/ale-bench-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "Vending-Bench 2 (Epoch mirror)",
              "rank": 12,
              "n_models": 66,
              "score": 9107.4583,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/vending-bench-2-external",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "APEX-Agents (Epoch mirror)",
              "rank": 6,
              "n_models": 56,
              "score": 0.683,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/apex-agents-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 122,
            "share": 0.04,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [
            {
              "tester": "epoch-ai",
              "benchmark": "GPQA Diamond",
              "rank": 9,
              "n_models": 295,
              "score": 0.9432,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/gpqa-diamond",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "SimpleQA Verified",
              "rank": 19,
              "n_models": 89,
              "score": 0.573,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/simpleqa-verified",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tiers 1-3 v2",
              "rank": 4,
              "n_models": 113,
              "score": 0.9088,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tiers-1-3-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "FrontierMath Tier 4 v2",
              "rank": 4,
              "n_models": 70,
              "score": 0.9268,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/frontiermath-tier-4-v2",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "OTIS Mock AIME 2024-2025",
              "rank": 1,
              "n_models": 274,
              "score": 1,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/otis-mock-aime-2024-2025",
              "licence": "display-ok"
            },
            {
              "tester": "epoch-ai",
              "benchmark": "CritPt (Epoch mirror) (best of 6 rows)",
              "rank": 15,
              "n_models": 170,
              "score": 0.28,
              "as_of": "2026-10-29",
              "url": "https://epoch.ai/benchmarks/critpt-external",
              "licence": "display-ok"
            }
          ],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "OpenAI model · 1M ctx · $1/$5"
    },
    {
      "id": "gpt-6-1-terra-pro",
      "name": "GPT-6.1 Terra Pro",
      "vendor": "OpenAI",
      "released": "2026-10-26",
      "context_window": 1050000,
      "price_input": 1,
      "price_output": 5,
      "speed_tps": null,
      "benchmarks": {
        "swe_bench": null,
        "gpqa": null,
        "aime": null,
        "mmlu_pro": null
      },
      "best_for": [],
      "strengths": [],
      "weaknesses": [],
      "verdict": null,
      "confidence": "low",
      "sources": [
        "https://openrouter.ai/openai/gpt-6.1-terra-pro"
      ],
      "coding_score": null,
      "coding_basis": null,
      "coding_confidence": "low",
      "use_well": [],
      "task_copy": {},
      "task_fit_judged": null,
      "usage": {
        "openrouter": {
          "category": "overall",
          "share": 0,
          "rank": 204,
          "as_of": "2026-10-28",
          "source_url": "https://openrouter.ai/api/frontend/v1/rankings/models"
        }
      },
      "auto_added": "2026-10-27",
      "availability": {
        "direct_api": null,
        "openrouter": true,
        "openrouter_id": "openai/gpt-6.1-terra-pro",
        "aws_bedrock": null,
        "google_vertex": null,
        "azure": null,
        "open_weights": null,
        "eu_hosting": null,
        "sources": [
          "https://openrouter.ai/api/v1/models"
        ]
      },
      "signals": {
        "coding": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "agents": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "bulk": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "writing": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "research": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "extraction": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "chat": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "vision": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "frontend": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        },
        "exec-summaries": {
          "usage_rank": null,
          "usage_share": null,
          "arena_rank": null,
          "expert_default": null,
          "families": 0
        }
      },
      "standings": {
        "as_of": "2026-10-29",
        "coding": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "agents": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "bulk": {
          "measured": [],
          "chosen": {
            "rank": 259,
            "share": 0,
            "tags": [],
            "n_models": 590,
            "as_of": "2026-10-23",
            "url": "https://openrouter.ai/rankings"
          },
          "preferred": null
        },
        "writing": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "research": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "extraction": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "chat": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "vision": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "frontend": {
          "measured": [],
          "chosen": null,
          "preferred": null
        },
        "exec-summaries": {
          "measured": [],
          "chosen": null,
          "preferred": null
        }
      },
      "task_fit": {
        "coding": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "agents": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model (agentic fit needs a coding baseline)"
        },
        "bulk": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "writing": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "research": {
          "score": null,
          "basis": [],
          "reason": "no GPQA score sourced for this model"
        },
        "extraction": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "chat": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        },
        "vision": {
          "score": null,
          "basis": [],
          "reason": "no \"vision\" tag on this model — no multimodal signal in the data"
        },
        "frontend": {
          "score": null,
          "basis": [],
          "reason": "no coding_score sourced for this model"
        },
        "exec-summaries": {
          "score": null,
          "basis": [],
          "reason": "no general-capability benchmark (gpqa/mmlu_pro) sourced for this model"
        }
      },
      "status": "ga",
      "adoption": "new",
      "best_for_line": "OpenAI model · 1M ctx · $1/$5"
    }
  ],
  "releases": [
    {
      "date": "2026-08-21",
      "vendor": "OpenAI",
      "title": "GPT-5.6 Sol price cut — $5/$30 → $4/$20 per 1M",
      "summary": "OpenAI cut Sol's API price by 'over 20%' ($5 → $4 input, $30 → $20 output per 1M tokens), promotional through at least Nov 21, 2026. Subscription usage unaffected.",
      "source": "https://community.openai.com/t/20-price-reduction-for-gpt-5-6-sol-api-codex-credits-and-chatgpt-work/1391726",
      "why": "If Sol is your top-tier pick, re-run your cost-per-task maths — every effort ladder on this page that plots Sol uses the old price.",
      "kind": "price"
    },
    {
      "date": "2026-08-13",
      "vendor": "Google",
      "title": "Google releases Gemini 3.7 Flash",
      "summary": "An algorithmic update three weeks after 3.6 Flash, billed as Google's 'most intelligent workhorse model yet' for coding, web dev and agents. Introductory price $0.75/$3.75 per 1M through Dec 31, 2026, then $1.50/$7.50. Stated gains over 3.6 Flash: FrontierCode 1.1 43.6% (from 34.4%), DeepSWE v1.1 65.3% (from 49.0%).",
      "source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/",
      "why": "The cheap-and-capable tier moved: at $0.75/$3.75 this is the price of a small model with mid-tier coding scores. The intro price doubles on Jan 1 — budget for that.",
      "kind": "model"
    },
    {
      "date": "2026-08-13",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4 Pro goes GA as '0813' — and gets peak-hour pricing",
      "summary": "V4 Pro leaves preview (1M context, 384K max output). Pricing now splits peak vs off-peak: output $1.98/1M off-peak vs $3.96 peak; cache-miss input $0.66 vs $1.32. Peak hours 01:00–04:00 and 06:00–10:00 UTC, Mon–Fri, effective Aug 16.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "why": "Overnight bulk jobs just got a clock: schedule them off-peak or pay double. The flat 'cheapest frontier-ish coder' story now depends on when you run.",
      "kind": "model"
    },
    {
      "date": "2026-08-12",
      "vendor": "xAI",
      "title": "xAI releases Grok 4.6",
      "summary": "Frontier multimodal model built on Grok 4.5, aimed at long-running agents, coding and visual work. 500K context; $2/$6 per 1M under 200K prompt tokens, $4/$12 above; four effort levels (low → xhigh). No benchmark numbers on xAI's own page.",
      "source": "https://docs.x.ai/developers/release-notes",
      "why": "On CursorBench it tops the board at a fraction of Opus-tier cost (see the effort ladders). Watch the 200K-token price cliff on long contexts.",
      "kind": "model"
    },
    {
      "date": "2026-07-30",
      "vendor": "OpenAI",
      "title": "GPT-5.6 Luna −80%, Terra −20%",
      "summary": "Luna drops from $1/$6 to $0.20/$1.20 per 1M; Terra from $2.50/$15 to $2/$12. Sol unchanged at the time (cut separately on Aug 21).",
      "source": "https://openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6/",
      "why": "Luna at $0.20/$1.20 is now a bulk-work candidate, not just a chat tier. Any ladder or chart costed before Jul 30 overstates Luna by 5x.",
      "kind": "price"
    },
    {
      "date": "2026-07-23",
      "vendor": "Anthropic",
      "title": "Anthropic releases Claude Opus 5",
      "summary": "Opus-tier pricing ($5/$25 per M, same as Opus 4.8) with Fable-class coding: 43.3% on Anthropic's Frontier-Bench v0.1 agentic-coding run vs Fable 5's 33.7% and Opus 4.8's 18.7%, and within 0.5% of Fable 5's peak CursorBench 3.2 score at max effort — at roughly half Fable's cost per task. 1M context, 128K output, five effort levels (low → max). Per-benchmark public numbers (SWE-bench Verified, GPQA) not published yet.",
      "source": "https://www.anthropic.com/news/claude-opus-5",
      "why": "If you pay for Claude: this is the new default for hard coding and agent work — Fable-level results at half the price. If you were paying Fable's $10/$50 for coding, check whether Opus 5 covers it.",
      "kind": "model"
    },
    {
      "date": "2026-07-16",
      "vendor": "Moonshot AI",
      "title": "Moonshot AI releases Kimi K3",
      "summary": "2.8T-param open-weight MoE (largest to date) at $3/$15 per M, 1M context, native vision. Tops the Frontend Code Arena (1679 Elo) ahead of Claude Fable 5, winning 6 of 7 frontend categories. Full weights due ~July 27; SWE-bench Verified not yet published.",
      "source": "https://www.tomshardware.com/tech-industry/artificial-intelligence/moonshot-releases-2-8-trillion-parameter-kimi-k3",
      "why": "If you build front-end or web UI: the new top pick for frontend coding — it edges out even Fable 5 in blind tests, at a fraction of the price ($3/$15).",
      "kind": "model"
    },
    {
      "date": "2026-07-09",
      "vendor": "OpenAI",
      "title": "GPT-5.6 (Sol/Terra/Luna) reaches general availability",
      "summary": "Three-tier family (Sol $5/$30, Terra $2.50/$15, Luna $1/$6) tops the Artificial Analysis Coding Index; previewed June 26 behind a US-government access list.",
      "source": "https://techcrunch.com/2026/07/09/openai-launches-its-new-family-of-models-with-gpt-5-6/",
      "why": "If you're on ChatGPT/OpenAI: three tiers now — Sol (top), Terra (value, ~$2.50/$15), Luna (cheap). Terra is the sweet spot for most.",
      "kind": "model"
    },
    {
      "date": "2026-07-08",
      "vendor": "xAI",
      "title": "xAI releases Grok 4.5",
      "summary": "'Opus-class' token-efficient coding/agentic model at $2/$6 per M — far cheaper than Opus 4.8 — bundled into Cursor; not available in EU at launch.",
      "source": "https://x.ai/news/grok-4-5",
      "why": "If you use Grok: the first version genuinely competitive for coding (86/100) at a low $2/$6.",
      "kind": "model"
    },
    {
      "date": "2026-07-01",
      "vendor": "Anthropic",
      "title": "Anthropic redeploys Claude Fable 5",
      "summary": "Fable 5 came back online after the US government lifted a June 12 export-control order that had pulled it offline — an AI-governance flashpoint.",
      "source": "https://llm-stats.com/llm-updates",
      "why": "Only matters for the hardest agent/reasoning runs — it's premium ($10/$50) and access-restricted; most people are better served by Opus or Sonnet.",
      "kind": "model"
    },
    {
      "date": "2026-06-30",
      "vendor": "Anthropic",
      "title": "Anthropic launches Claude Sonnet 5",
      "summary": "Cheaper agent-focused model, near-Opus quality; intro pricing $2/$10 per M through Aug 31 (then $3/$15), 1M context, default on Free/Pro.",
      "source": "https://techcrunch.com/2026/06/30/anthropic-launches-claude-sonnet-5-as-a-cheaper-way-to-run-agents/",
      "why": "Big one for coders: ~85 coding at $2/$10 — the best coding value right now. (Intro price; rises to $3/$15 on 2026-09-01.)",
      "kind": "model"
    },
    {
      "date": "2026-06-20",
      "vendor": "DeepSeek / Z.ai / Alibaba",
      "title": "Open-weight wave: DeepSeek V4 preview, GLM-5.2, Qwen3-Coder-Next",
      "summary": "A cluster of 1M-token open-weight coding models landed, pressuring closed-model pricing — including the compact 80B/3B-active Qwen3-Coder-Next.",
      "source": "https://kingy.ai/news/best-open-weight-ai-models-in-2026-glm-5-2-vs-deepseek-v4-vs-kimi-k2-6-vs-qwen-vs-mistral/",
      "why": "Care if you self-host or want low cost: strong open-weight options (DeepSeek V4, Qwen3-Coder) at a fraction of frontier prices.",
      "kind": "model"
    },
    {
      "date": "2026-06-12",
      "vendor": "Moonshot AI",
      "title": "Moonshot ships Kimi K2.7 Code",
      "summary": "Open-weight coding model claiming +21.8% over K2.6 on its own Kimi Code Bench v2 — continued fast iteration in Chinese open-weight coders.",
      "source": "https://kingy.ai/news/best-open-weight-ai-models-in-2026-glm-5-2-vs-deepseek-v4-vs-kimi-k2-6-vs-qwen-vs-mistral/",
      "why": "Niche: a cheap, coding-focused open model — worth a look if you want low-cost agentic coding.",
      "kind": "model"
    },
    {
      "date": "2026-05-19",
      "vendor": "Google",
      "title": "Google ships Gemini 3.5 Flash",
      "summary": "Fast Flash-tier model with near-Pro reasoning (GPQA ~92%), 1M context, ~278 tok/s, at $1.50/$9 per M.",
      "source": "https://ai.google.dev/gemini-api/docs/pricing",
      "why": "If you use Gemini: a fast, cheap workhorse (coding 83 at ~$1.50/$9).",
      "kind": "model"
    },
    {
      "date": "2026-05-28",
      "vendor": "Anthropic",
      "title": "Claude Opus 4.8",
      "summary": "Flagship Opus tier at $5/$25 per M, 1M context, 88.6% SWE-bench Verified — the strong coding/agentic default below Fable's price.",
      "source": "https://www.anthropic.com/news/claude-opus-4-8",
      "why": "Anthropic's flagship you'll actually reach for — top-tier coding and reasoning at $5/$25.",
      "kind": "model"
    },
    {
      "date": "2026-04-24",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4-Flash and V4-Pro",
      "summary": "Open-weight (MIT) 1M-context models at rock-bottom prices; V4-Pro hits ~80.6% SWE-bench Verified with near-free cache-hit input.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "why": "The value story: near-frontier coding at ~$0.43/$0.87 (Pro) — big savings when that quality is enough.",
      "kind": "model"
    },
    {
      "date": "2026-04-23",
      "vendor": "OpenAI",
      "title": "OpenAI introduces GPT-5.5",
      "summary": "Prior flagship at $5/$30 per M, 1M context, 81.9% SWE-bench Verified and LMArena Elo 1475 — now largely superseded by GPT-5.6 at equal cost.",
      "source": "https://openai.com/index/introducing-gpt-5-5/",
      "why": "Superseded by GPT-5.6 — only relevant if something is still pinned to it.",
      "kind": "model"
    },
    {
      "date": "2026-04-20",
      "vendor": "Moonshot AI",
      "title": "Moonshot AI releases Kimi K2.6",
      "summary": "1T-param MoE (32B active) open-weight agentic coder at 80.2% SWE-bench Verified, built for long-horizon multi-step tasks.",
      "source": "https://platform.kimi.ai/docs/pricing/chat-k26",
      "why": "A cheap, long-context open model — niche unless you want low-cost bulk work.",
      "kind": "model"
    },
    {
      "date": "2025-10-01",
      "vendor": "Anthropic",
      "title": "Claude Haiku 4.5",
      "summary": "Cheapest/fastest Claude at $1/$5 per M, 200K context — built for high-volume classification, routing, and extraction.",
      "source": "https://www.anthropic.com/news/claude-haiku-4-5",
      "why": "Anthropic's cheap/fast tier ($1/$5) — a good default for high-volume simple tasks.",
      "kind": "model"
    },
    {
      "kind": "model",
      "date": "2026-05-21",
      "vendor": "Tencent",
      "title": "Tencent releases Hy-MT2-1.8B",
      "summary": "Tencent Hunyuan open-sourced the Hy-MT2 translation family (1.8B/7B/30B-A3B) under Apache-2.0, using 1.25-bit quantization to fit the 1.8B model into 440MB for on-device use.",
      "source": "https://huggingface.co/tencent/Hy-MT2-1.8B",
      "why": "Extends Tencent's open translation lineup past Hy-MT1.5, with claimed gains over larger commercial and open models on multilingual translation tasks."
    },
    {
      "kind": "model",
      "date": "2026-05-21",
      "vendor": "Tencent",
      "title": "Tencent releases Hy-MT2-30B-A3B",
      "summary": "Tencent Hunyuan open-sourced the Hy-MT2 translation family (1.8B/7B/30B-A3B) under Apache-2.0; the 30B-A3B variant is a 30B-total/3B-active MoE aimed at professional-grade translation across 33 languages.",
      "source": "https://github.com/Tencent-Hunyuan/Hy-MT2",
      "why": "Extends Tencent's open translation lineup past Hy-MT1.5, with claimed gains over larger commercial and open models on multilingual translation tasks."
    },
    {
      "kind": "model",
      "date": "2026-08-14",
      "vendor": "Z.ai (Zhipu)",
      "title": "Z.ai (Zhipu) releases GLM-5.3",
      "summary": "Z.ai shipped GLM-5.3, a post-training upgrade of the GLM-5.2 base aimed at coding and cyber-defense work, lifting Terminal-Bench 3.0 from 4.6% to 28.3%.",
      "source": "https://x.com/Zai_org/status/2088132965922476159",
      "why": "Available now via Z.ai's GLM Coding Plan and ZCode; broader API access and open weights are staged to follow after safety hardening."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4-Flash input price rise — $0.14 → $0.44 per 1M (+214%)",
      "summary": "DeepSeek now lists DeepSeek V4-Flash at $0.44 per 1M input tokens, from $0.14. Confirmed by two independent price feeds.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "why": "Budgets built on the old DeepSeek V4-Flash price are now 214% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4-Flash output price rise — $0.28 → $1.32 per 1M (+371%)",
      "summary": "DeepSeek now lists DeepSeek V4-Flash at $1.32 per 1M output tokens, from $0.28. Confirmed by two independent price feeds.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "why": "Budgets built on the old DeepSeek V4-Flash price are now 371% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4-Pro input price rise — $0.43 → $1.32 per 1M (+203%)",
      "summary": "DeepSeek now lists DeepSeek V4-Pro at $1.32 per 1M input tokens, from $0.43. Confirmed by two independent price feeds.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "why": "Budgets built on the old DeepSeek V4-Pro price are now 203% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4-Pro output price rise — $0.87 → $3.96 per 1M (+355%)",
      "summary": "DeepSeek now lists DeepSeek V4-Pro at $3.96 per 1M output tokens, from $0.87. Confirmed by two independent price feeds.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "why": "Budgets built on the old DeepSeek V4-Pro price are now 355% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "Google",
      "title": "Google: Gemini 3.7 Flash input price rise — $0.38 → $0.75 per 1M (+97%)",
      "summary": "google now lists Google: Gemini 3.7 Flash at $0.75 per 1M input tokens, from $0.38. Confirmed by two independent price feeds.",
      "source": "https://ai.google.dev/gemini-api/docs/pricing",
      "why": "Budgets built on the old Google: Gemini 3.7 Flash price are now 97% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "OpenAI",
      "title": "GPT-5.6 Sol input price cut — $5 → $4 per 1M (-20%)",
      "summary": "OpenAI now lists GPT-5.6 Sol at $4 per 1M input tokens, from $5. Confirmed by two independent price feeds.",
      "source": "https://developers.openai.com/api/docs/pricing",
      "why": "Re-run any cost-per-task maths that used GPT-5.6 Sol — the old number overstates it by 20%."
    },
    {
      "kind": "price",
      "date": "2026-08-25",
      "vendor": "OpenAI",
      "title": "GPT-5.6 Sol output price cut — $30 → $20 per 1M (-33%)",
      "summary": "OpenAI now lists GPT-5.6 Sol at $20 per 1M output tokens, from $30. Confirmed by two independent price feeds.",
      "source": "https://developers.openai.com/api/docs/pricing",
      "why": "Re-run any cost-per-task maths that used GPT-5.6 Sol — the old number overstates it by 33%."
    },
    {
      "kind": "model",
      "date": "2026-08-03",
      "vendor": "Alibaba (Qwen)",
      "title": "Qwen releases Qwen3.8 Max",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/qwen/qwen3.8-max",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-08-26",
      "vendor": "Alibaba (Qwen)",
      "title": "Qwen releases Qwen3.8 Flash",
      "summary": "Alibaba released Qwen3.8-Flash, a 125B-total/6B-active MoE model previewing the hybrid-attention architecture behind Qwen4, with a 262K native context extensible to 1M via YaRN.",
      "source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
      "why": "Positioned as a cost-efficiency play — Alibaba says it beats Qwen3.7-Plus at roughly one-ninth the training cost, with the biggest gains in coding and office tasks."
    },
    {
      "kind": "model",
      "date": "2026-08-28",
      "vendor": "Tencent",
      "title": "Tencent releases Hy4 preview",
      "summary": "Tencent Hunyuan open-sourced Hy4 preview, a 770B-total/49B-active MoE flagship under Apache-2.0 with a 1M-token context, co-designed with Tencent's CodeBuddy and WorkBuddy products.",
      "source": "https://www.tencent.com/tencent-releases-and-open-sources-tencent-hy4-preview/",
      "why": "Positioned as Tencent's new flagship for coding, office, and agentic productivity tasks; Tencent's internal blind evaluation reportedly edged out GLM 5.3 and Kimi K3 on engineering tasks."
    },
    {
      "kind": "model",
      "date": "2026-05-21",
      "vendor": "Tencent",
      "title": "Tencent releases Hy-MT2-7B",
      "summary": "Tencent Hunyuan open-sourced the Hy-MT2 translation family (1.8B/7B/30B-A3B) under Apache-2.0; the 7B variant targets balanced performance across 33 languages.",
      "source": "https://huggingface.co/tencent/Hy-MT2-7B",
      "why": "Fills out the Hy-MT2 family already in the catalog (1.8B and 30B-A3B) with the middle 7B tier."
    },
    {
      "kind": "price",
      "date": "2026-09-01",
      "vendor": "Google",
      "title": "Google: Gemini 3.7 Flash output price rise — $1.88 → $3.75 per 1M (+99%)",
      "summary": "google now lists Google: Gemini 3.7 Flash at $3.75 per 1M output tokens, from $1.88. Confirmed by two independent price feeds.",
      "source": "https://ai.google.dev/gemini-api/docs/pricing",
      "why": "Budgets built on the old Google: Gemini 3.7 Flash price are now 99% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "model",
      "date": "2026-08-25",
      "vendor": "IBM",
      "title": "IBM releases Granite 4.2 8B",
      "summary": "IBM released Granite 4.2 on Aug 25, 2026: open (Apache 2.0) 3B/8B/30B dense reasoning models with a per-request thinking toggle.",
      "source": "https://research.ibm.com/blog/introducing-granite-4-2",
      "why": "Adds a native reasoning switch plus an added agentic-RL training stage (tool-calling, multi-step coding, terminal use) to IBM's open Granite line."
    },
    {
      "kind": "model",
      "date": "2026-09-02",
      "vendor": "Meta",
      "title": "Meta releases Muse Spark 1.3 Contributor",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/meta/muse-spark-1.3-contributor",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-02",
      "vendor": "Meta",
      "title": "Meta releases Muse Spark 1.3",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/meta/muse-spark-1.3",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-02",
      "vendor": "Google",
      "title": "Google releases Gemini 3.8 Flash",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/google/gemini-3.8-flash",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-04",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6 Astra",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6-astra",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-04",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6 Astra Pro",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6-astra-pro",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-01",
      "vendor": "Anthropic",
      "title": "Anthropic releases Claude Fable 5.1",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/anthropic/claude-fable-5.1",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-08-21",
      "vendor": "Meta",
      "title": "Meta releases Muse Spark 1.2 Contributor",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/meta/muse-spark-1.2-contributor",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-08-21",
      "vendor": "DeepSeek",
      "title": "DeepSeek releases DeepSeek V4 Flash Vision Exp",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/deepseek/deepseek-v4-flash-vision-exp",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-08-12",
      "vendor": "DeepSeek",
      "title": "DeepSeek releases DeepSeek V4 Pro 0813",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/deepseek/deepseek-v4-pro-0813",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-07-31",
      "vendor": "DeepSeek",
      "title": "DeepSeek releases DeepSeek V4 Flash 0731",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/deepseek/deepseek-v4-flash-0731",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "price",
      "date": "2026-09-07",
      "vendor": "Poolside",
      "title": "Laguna XS 2.1 input price cut — $0.10 → $0.06 per 1M (-40%)",
      "summary": "Poolside now lists Laguna XS 2.1 at $0.06 per 1M input tokens, from $0.10. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Re-run any cost-per-task maths that used Laguna XS 2.1 — the old number overstates it by 40%."
    },
    {
      "kind": "price",
      "date": "2026-09-07",
      "vendor": "Poolside",
      "title": "Laguna XS 2.1 output price cut — $0.20 → $0.12 per 1M (-40%)",
      "summary": "Poolside now lists Laguna XS 2.1 at $0.12 per 1M output tokens, from $0.20. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Re-run any cost-per-task maths that used Laguna XS 2.1 — the old number overstates it by 40%."
    },
    {
      "kind": "price",
      "date": "2026-09-07",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4 Flash Vision Exp input price rise — $0.22 → $0.44 per 1M (+100%)",
      "summary": "DeepSeek now lists DeepSeek V4 Flash Vision Exp at $0.44 per 1M input tokens, from $0.22. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Budgets built on the old DeepSeek V4 Flash Vision Exp price are now 100% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-09-07",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4 Flash Vision Exp output price rise — $0.66 → $1.32 per 1M (+100%)",
      "summary": "DeepSeek now lists DeepSeek V4 Flash Vision Exp at $1.32 per 1M output tokens, from $0.66. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Budgets built on the old DeepSeek V4 Flash Vision Exp price are now 100% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "model",
      "date": "2026-09-10",
      "vendor": "DeepSeek",
      "title": "DeepSeek releases DeepSeek V4.1 Flash",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/deepseek/deepseek-v4.1-flash",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "price",
      "date": "2026-09-18",
      "vendor": "Upstage",
      "title": "Solar Pro 4 input price cut — $0.30 → $0.09 per 1M (-70%)",
      "summary": "Upstage now lists Solar Pro 4 at $0.09 per 1M input tokens, from $0.30. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Re-run any cost-per-task maths that used Solar Pro 4 — the old number overstates it by 70%."
    },
    {
      "kind": "price",
      "date": "2026-09-18",
      "vendor": "Upstage",
      "title": "Solar Pro 4 output price cut — $1.20 → $0.36 per 1M (-70%)",
      "summary": "Upstage now lists Solar Pro 4 at $0.36 per 1M output tokens, from $1.20. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Re-run any cost-per-task maths that used Solar Pro 4 — the old number overstates it by 70%."
    },
    {
      "kind": "model",
      "date": "2026-09-21",
      "vendor": "xAI",
      "title": "xAI releases Grok 4.7",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/x-ai/grok-4.7",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-22",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6 Luna Pro",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6-luna-pro",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-22",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6 Luna",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6-luna",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-22",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6 Sol Pro",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6-sol-pro",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-22",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6 Sol",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6-sol",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-22",
      "vendor": "Anthropic",
      "title": "Anthropic releases Claude Opus 5.5",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/anthropic/claude-opus-5.5",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-21",
      "vendor": "Alibaba (Qwen)",
      "title": "Alibaba (Qwen) releases Qwen3.8 Omni Flash",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/qwen/qwen3.8-omni-flash",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-22",
      "vendor": "Cohere",
      "title": "Cohere releases Command A+",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/cohere/command-a-plus",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-23",
      "vendor": "Alibaba (Qwen)",
      "title": "Alibaba (Qwen) releases Qwen3.8 Max Prime",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/qwen/qwen3.8-max-prime",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "price",
      "date": "2026-09-28",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4.1 Flash input price rise — $0.15 → $0.30 per 1M (+100%)",
      "summary": "DeepSeek now lists DeepSeek V4.1 Flash at $0.30 per 1M input tokens, from $0.15. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Budgets built on the old DeepSeek V4.1 Flash price are now 100% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-09-28",
      "vendor": "DeepSeek",
      "title": "DeepSeek V4.1 Flash output price rise — $0.60 → $1.20 per 1M (+100%)",
      "summary": "DeepSeek now lists DeepSeek V4.1 Flash at $1.20 per 1M output tokens, from $0.60. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Budgets built on the old DeepSeek V4.1 Flash price are now 100% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-09-28",
      "vendor": "xAI",
      "title": "Grok 4.7 input price rise — $1.60 → $2 per 1M (+25%)",
      "summary": "xAI now lists Grok 4.7 at $2 per 1M input tokens, from $1.60. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Budgets built on the old Grok 4.7 price are now 25% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "price",
      "date": "2026-09-28",
      "vendor": "xAI",
      "title": "Grok 4.7 output price rise — $4.80 → $6 per 1M (+25%)",
      "summary": "xAI now lists Grok 4.7 at $6 per 1M output tokens, from $4.80. Confirmed by two independent price feeds.",
      "source": "https://openrouter.ai/api/v1/models",
      "why": "Budgets built on the old Grok 4.7 price are now 25% short. Check whether a cheaper tier covers the job."
    },
    {
      "kind": "model",
      "date": "2026-09-28",
      "vendor": "Anthropic",
      "title": "Anthropic releases Claude Sonnet 5.5",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/anthropic/claude-sonnet-5.5",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-29",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6.1 Sol Pro",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6.1-sol-pro",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-29",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6.1 Sol",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6.1-sol",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-06",
      "vendor": "Mistral AI",
      "title": "Mistral AI releases Mistral Large 4",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/mistralai/mistral-large-4-0",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-07",
      "vendor": "Anthropic",
      "title": "Anthropic releases Claude Haiku 5.5",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/anthropic/claude-haiku-5.5",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-13",
      "vendor": "Google",
      "title": "Google releases Gemini 4 Argon",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/google/gemini-4-argon",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-17",
      "vendor": "DeepSeek",
      "title": "DeepSeek releases DeepSeek V4.1 Flash 1017",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/deepseek/deepseek-v4.1-flash-1017",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-15",
      "vendor": "Meta",
      "title": "Meta releases Muse Spark Open",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/meta/muse-spark-open",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-15",
      "vendor": "NVIDIA",
      "title": "NVIDIA releases Nemotron 3.5 Super VL",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/nvidia/nemotron-3.5-super-vl",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-12",
      "vendor": "Meta",
      "title": "Meta releases Muse Wonder",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/meta/muse-wonder",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-02",
      "vendor": "InclusionAI (Ant Group)",
      "title": "InclusionAI (Ant Group) releases Ling 3.1 Flash",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/inclusionai/ling-3.1-flash",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-23",
      "vendor": "Z.ai (Zhipu)",
      "title": "Z.ai (Zhipu) releases GLM 5.3 Prime",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/z-ai/glm-5.3-prime",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-23",
      "vendor": "AionLabs",
      "title": "AionLabs releases Aion 3.5 Mini",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/aion-labs/aion-3.5-mini",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-23",
      "vendor": "AionLabs",
      "title": "AionLabs releases Aion 3.5",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/aion-labs/aion-3.5",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-23",
      "vendor": "Upstage",
      "title": "Upstage releases Solar Mini 4",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/upstage/solar-mini4",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-18",
      "vendor": "Z.ai (Zhipu)",
      "title": "Z.ai (Zhipu) releases GLM 5.3 FlashX",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/z-ai/glm-5.3-flashx",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-11",
      "vendor": "Sakana AI",
      "title": "Sakana AI releases Fugu Ultra v2",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/sakana/fugu-ultra-v2",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-11",
      "vendor": "Sakana AI",
      "title": "Sakana AI releases Fugu Max",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/sakana/fugu-max",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-09-10",
      "vendor": "InclusionAI (Ant Group)",
      "title": "InclusionAI (Ant Group) releases Ling 3.0 Flash VL",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/inclusionai/ling-3.0-flash-vl",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-23",
      "vendor": "Moonshot AI",
      "title": "Moonshot AI releases Kimi K3.1 Pro",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/moonshotai/kimi-k3.1-pro",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-26",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6.1 Terra",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6.1-terra",
      "why": "Priced and available now; treat capability as unproven until scores land."
    },
    {
      "kind": "model",
      "date": "2026-10-26",
      "vendor": "OpenAI",
      "title": "OpenAI releases GPT-6.1 Terra Pro",
      "summary": "Listed with sourced pricing and context window; benchmark scores are pending publication.",
      "source": "https://openrouter.ai/openai/gpt-6.1-terra-pro",
      "why": "Priced and available now; treat capability as unproven until scores land."
    }
  ],
  "benchmarks_legend": {
    "swe_bench": "SWE-bench Verified — % of real-world GitHub issues resolved end-to-end (agentic coding). Note: several 2026 models published only SWE-bench Pro, a different/harder variant, so their Verified cell is null.",
    "gpqa": "GPQA Diamond — % correct on graduate-level, Google-proof science reasoning questions. Figures sourced from Epoch AI are independently run and reported at the reasoning-effort setting Epoch used (max for Opus 5, Kimi K3, GPT-5.6 Terra and Luna; xhigh for Sonnet 5; high for Grok 4.5), so they are best-case settings rather than defaults.",
    "aime": "AIME — % on advanced high-school competition math (American Invitational Mathematics Examination).",
    "mmlu_pro": "MMLU-Pro — % on a harder, multi-discipline knowledge and reasoning benchmark."
  },
  "notes": "Pricing is verified against official vendor pricing pages (standard tier, USD per 1M tokens) — treat it as the reliable layer. Benchmarks are directional: they mix primary leaderboards, Epoch AI's independently-run figures, and third-party aggregators, and a confidence flag rides on every model. Claude Opus 5 (released 2026-07-23) is priced at Anthropic's published $5/$25; its GPQA (93.9) is Epoch's independent run, but SWE-bench Verified is still unpublished, so its coding score remains an agentic-suite-led estimate at medium confidence. Kimi K3 is scored on its frontend-coding lead (#1 in the Frontend Code Arena); its GPQA (93.1) is Epoch-run, SWE-bench Verified is still unpublished, and its open-weight release (slated 2026-07-27) has not been re-verified as shipped here — treat self-hosting as unconfirmed. Several 2026 models publish only agentic evals that don't map to SWE-bench Verified; those cells are left blank rather than cross-filed. A blank (—) means 'not reliably sourced,' never a guess. Sonnet 5 shows its current intro price ($2/$10; rises to $3/$15 on Sept 1, 2026). Re-verify any single benchmark against a primary leaderboard before betting real spend on it.",
  "coding_score_note": "Coding score (0–100) is a unified, comparable coding-ability estimate. Where an official SWE-bench Verified % exists it IS that number (high confidence). Where it doesn't — several strong 2026 models ship before formal benchmarks — it's estimated from sourced signals (Artificial Analysis Coding Index, LMArena Code Elo, SWE-bench Pro, LiveCodeBench) calibrated against the benchmarked models (medium confidence), or expert consensus (low). Each row shows its basis and confidence. These signals sometimes disagree — treat medium/low scores as directional.",
  "use_well_note": "use_well = practical, plain-English guidance on getting the most from a model you already have. Derived from the model's sourced pricing, context, cache and mode facts above (plus vendor-documented features like Claude's adaptive thinking) — guidance, not benchmarks. Same independence rule: factual, cost-first, never promotional.",
  "effort_ladders": [
    {
      "id": "cursorbench-agentic-coding",
      "suite": "CursorBench",
      "task": "Agentic coding",
      "x_label": "Cost per task (USD, log scale)",
      "y_label": "Score (%)",
      "as_of": "2026-10-28",
      "publisher": "Epoch AI (mirroring Cursor's CursorBench)",
      "source_kind": "third-party",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "confidence": "high",
      "method": "Exact published values from Epoch AI's CC-BY benchmark export (cursorbench_external.csv in benchmark_data.zip), which mirrors Cursor's own CursorBench leaderboard. The model list is rebuilt on every data refresh by one rule: a model is shown when it is in this catalog and the file measures it at 3 or more effort settings, each with both a cost and a score. Only the newest model in each line is kept (a line is the model name without its version numbers, so version 5.5 of a model replaces version 5), and a model released more than 150 days ago drops off. Models are listed by how much they are used on OpenRouter, most used first. Score is the share of tasks solved, as a percentage; cost is Cursor's cost per task, from the tokens each model used at its published per-million prices. A setting missing its cost or its score is left out, never filled in.",
      "harness": "Cursor's agent running CursorBench 3.2 (instruction-following and advanced tool-use problems added Jul 8 2026), one entry per reasoning-effort setting. Cursor notes results are subject to variance and small score differences may not be meaningful.",
      "caveat": "Independent of Anthropic, but not of everyone: CursorBench is run and published by Cursor, which ships its own coding model (Composer), so it is a benchmark by a company with a horse in the coding race. Epoch AI mirrors those numbers under CC-BY, it does not re-run them — and Epoch has its own disclosed conflict elsewhere, having had OpenAI fund the FrontierMath benchmark with dataset access, which it disclosed only after the fact. Read this as a second opinion measured on a different task by a different party, not as a neutral referee settling the question.",
      "levels": [
        "minimal",
        "low",
        "medium",
        "high",
        "xhigh",
        "max"
      ],
      "default_count": 5,
      "series": [
        {
          "model_id": "claude-opus-5-5",
          "label": "Claude Opus 5.5",
          "source_key": "claude-opus-5-5",
          "usage_rank": 6,
          "points": [
            {
              "effort": "low",
              "cost": 1.17,
              "score": 43.7
            },
            {
              "effort": "medium",
              "cost": 2.91,
              "score": 52.5
            },
            {
              "effort": "high",
              "cost": 3.97,
              "score": 56
            },
            {
              "effort": "xhigh",
              "cost": 6.98,
              "score": 56
            },
            {
              "effort": "max",
              "cost": 13.43,
              "score": 57.8
            }
          ]
        },
        {
          "model_id": "glm-5-3",
          "label": "GLM-5.3",
          "source_key": "glm-5.3",
          "usage_rank": 9,
          "points": [
            {
              "effort": "low",
              "cost": 2.04,
              "score": 33.3
            },
            {
              "effort": "high",
              "cost": 3.24,
              "score": 38
            },
            {
              "effort": "max",
              "cost": 5.05,
              "score": 42.6
            }
          ]
        },
        {
          "model_id": "claude-haiku-5-5",
          "label": "Claude Haiku 5.5",
          "source_key": "claude-haiku-5-5",
          "usage_rank": 16,
          "points": [
            {
              "effort": "low",
              "cost": 0.08,
              "score": 30.9
            },
            {
              "effort": "medium",
              "cost": 0.17,
              "score": 36.9
            },
            {
              "effort": "high",
              "cost": 0.32,
              "score": 42.3
            },
            {
              "effort": "xhigh",
              "cost": 0.56,
              "score": 44.3
            },
            {
              "effort": "max",
              "cost": 1.12,
              "score": 48.4
            }
          ]
        },
        {
          "model_id": "claude-sonnet-5-5",
          "label": "Claude Sonnet 5.5",
          "source_key": "claude-sonnet-5-5",
          "usage_rank": 19,
          "points": [
            {
              "effort": "low",
              "cost": 0.37,
              "score": 35.8
            },
            {
              "effort": "medium",
              "cost": 0.52,
              "score": 39.2
            },
            {
              "effort": "high",
              "cost": 1.2,
              "score": 47.8
            },
            {
              "effort": "xhigh",
              "cost": 2.81,
              "score": 53.1
            },
            {
              "effort": "max",
              "cost": 7.05,
              "score": 55.5
            }
          ]
        },
        {
          "model_id": "gpt-5-6-luna",
          "label": "GPT-5.6 Luna",
          "source_key": "gpt-5.6-luna",
          "usage_rank": 20,
          "points": [
            {
              "effort": "low",
              "cost": 0.03,
              "score": 16
            },
            {
              "effort": "medium",
              "cost": 0.08,
              "score": 22.2
            },
            {
              "effort": "high",
              "cost": 0.25,
              "score": 29.4
            },
            {
              "effort": "xhigh",
              "cost": 0.44,
              "score": 33
            },
            {
              "effort": "max",
              "cost": 1.03,
              "score": 35.9
            }
          ]
        },
        {
          "model_id": "gemini-4-argon",
          "label": "Gemini 4 Argon",
          "source_key": "gemini-4-argon",
          "usage_rank": 32,
          "points": [
            {
              "effort": "minimal",
              "cost": 3.54,
              "score": 48.8
            },
            {
              "effort": "low",
              "cost": 4.09,
              "score": 50.6
            },
            {
              "effort": "medium",
              "cost": 4.39,
              "score": 52.8
            },
            {
              "effort": "high",
              "cost": 4.57,
              "score": 52.9
            }
          ]
        },
        {
          "model_id": "grok-4-7",
          "label": "Grok 4.7",
          "source_key": "grok-4.7",
          "usage_rank": 41,
          "points": [
            {
              "effort": "low",
              "cost": 1.58,
              "score": 33.1
            },
            {
              "effort": "medium",
              "cost": 3.49,
              "score": 41.6
            },
            {
              "effort": "high",
              "cost": 4.69,
              "score": 43.9
            },
            {
              "effort": "xhigh",
              "cost": 6.01,
              "score": 46.3
            }
          ]
        },
        {
          "model_id": "claude-fable-5-1",
          "label": "Claude Fable 5.1",
          "source_key": "claude-fable-5-1",
          "usage_rank": 51,
          "points": [
            {
              "effort": "low",
              "cost": 5.44,
              "score": 45.1
            },
            {
              "effort": "medium",
              "cost": 7.05,
              "score": 46.8
            },
            {
              "effort": "high",
              "cost": 9.08,
              "score": 49.2
            },
            {
              "effort": "xhigh",
              "cost": 13.01,
              "score": 51.6
            },
            {
              "effort": "max",
              "cost": 17.28,
              "score": 51.8
            }
          ]
        },
        {
          "model_id": "gpt-5-6-sol",
          "label": "GPT-5.6 Sol",
          "source_key": "gpt-5.6-sol",
          "usage_rank": 55,
          "points": [
            {
              "effort": "low",
              "cost": 0.87,
              "score": 24.6
            },
            {
              "effort": "medium",
              "cost": 1.77,
              "score": 31.1
            },
            {
              "effort": "high",
              "cost": 2.85,
              "score": 35.7
            },
            {
              "effort": "xhigh",
              "cost": 4.4,
              "score": 37.7
            },
            {
              "effort": "max",
              "cost": 8.23,
              "score": 41.7
            }
          ]
        },
        {
          "model_id": "gpt-5-6-terra",
          "label": "GPT-5.6 Terra",
          "source_key": "gpt-5.6-terra",
          "usage_rank": 63,
          "points": [
            {
              "effort": "low",
              "cost": 0.52,
              "score": 25.2
            },
            {
              "effort": "medium",
              "cost": 0.64,
              "score": 27.6
            },
            {
              "effort": "high",
              "cost": 1.11,
              "score": 30.7
            },
            {
              "effort": "xhigh",
              "cost": 1.81,
              "score": 33.6
            },
            {
              "effort": "max",
              "cost": 5.14,
              "score": 41.3
            }
          ]
        },
        {
          "model_id": "muse-spark-1-3",
          "label": "Muse Spark 1.3",
          "source_key": "muse-spark-1.3",
          "usage_rank": 96,
          "points": [
            {
              "effort": "minimal",
              "cost": 0.56,
              "score": 24.3
            },
            {
              "effort": "low",
              "cost": 0.93,
              "score": 29.3
            },
            {
              "effort": "medium",
              "cost": 1.49,
              "score": 32.6
            },
            {
              "effort": "high",
              "cost": 1.66,
              "score": 33.4
            },
            {
              "effort": "xhigh",
              "cost": 2.1,
              "score": 37.5
            },
            {
              "effort": "max",
              "cost": 2.64,
              "score": 41.6
            }
          ]
        },
        {
          "model_id": "deepseek-v4-1-flash-1017",
          "label": "DeepSeek V4.1 Flash 1017",
          "source_key": "deepseek-v4.1-flash-1017",
          "usage_rank": null,
          "points": [
            {
              "effort": "low",
              "cost": 0.6,
              "score": 44.9
            },
            {
              "effort": "medium",
              "cost": 0.8,
              "score": 46.3
            },
            {
              "effort": "high",
              "cost": 0.77,
              "score": 45.6
            },
            {
              "effort": "max",
              "cost": 0.98,
              "score": 47.7
            }
          ]
        }
      ]
    },
    {
      "id": "frontier-bench-v0-1-agentic-coding",
      "suite": "Frontier-Bench v0.1",
      "task": "Agentic coding",
      "x_label": "Cost per attempt (USD, log scale)",
      "y_label": "Score (%)",
      "as_of": "2026-07-24",
      "publisher": "Anthropic",
      "source_kind": "vendor-reported",
      "source": "https://www.anthropic.com/news/claude-opus-5",
      "confidence": "medium",
      "method": "Values digitised from the published figure in Anthropic's Claude Opus 5 announcement — point positions read off the chart, so treat costs as ±~5% and scores as ±~0.5 points. The max-effort endpoints are exact: 43.3 / 33.7 / 18.7 are quoted numerically in the launch coverage and match the plotted points, which is what anchors the rest of the reading.",
      "harness": "Internal run of Frontier-Bench v0.1 on the mini-SWE-agent harness with a GKE backend; mean reward over 5 attempts per task. Opus 4.8 served as fallback on safety-classifier refusals for Opus 5 and Fable 5.",
      "caveat": "Anthropic's own benchmark, own harness, own run — and three of the four models plotted are its own. Read it for the SHAPE of each model's cost/quality curve (where extra spend stops buying accuracy), not as a settled cross-lab ranking. No third party has reproduced it.",
      "levels": [
        "low",
        "medium",
        "high",
        "xhigh",
        "max"
      ],
      "series": [
        {
          "model_id": "claude-opus-5",
          "label": "Opus 5",
          "color": "#e05c3e",
          "points": [
            {
              "effort": "low",
              "cost": 5.63,
              "score": 25.8
            },
            {
              "effort": "medium",
              "cost": 8.2,
              "score": 35
            },
            {
              "effort": "high",
              "cost": 11.14,
              "score": 39.5
            },
            {
              "effort": "xhigh",
              "cost": 14.18,
              "score": 44.3
            },
            {
              "effort": "max",
              "cost": 16.48,
              "score": 43.3
            }
          ]
        },
        {
          "model_id": "claude-fable-5",
          "label": "Fable 5",
          "color": "#e8a93a",
          "points": [
            {
              "effort": "low",
              "cost": 10.17,
              "score": 18.1
            },
            {
              "effort": "medium",
              "cost": 11.5,
              "score": 25
            },
            {
              "effort": "high",
              "cost": 14.18,
              "score": 29.1
            },
            {
              "effort": "xhigh",
              "cost": 18.35,
              "score": 31.6
            },
            {
              "effort": "max",
              "cost": 28.2,
              "score": 33.7
            }
          ]
        },
        {
          "model_id": "claude-opus-4-8",
          "label": "Opus 4.8",
          "color": "#3f7fe0",
          "points": [
            {
              "effort": "low",
              "cost": 4.74,
              "score": 6.7
            },
            {
              "effort": "medium",
              "cost": 7.17,
              "score": 9.7
            },
            {
              "effort": "high",
              "cost": 8.52,
              "score": 12.9
            },
            {
              "effort": "xhigh",
              "cost": 11.02,
              "score": 15.5
            },
            {
              "effort": "max",
              "cost": 15.62,
              "score": 18.7
            }
          ]
        },
        {
          "model_id": "gpt-5-6-sol",
          "label": "GPT-5.6 Sol",
          "color": "#9aa0ab",
          "level_note": "OpenAI exposes its own reasoning-effort settings, not Anthropic's five names — these are the rungs as Anthropic plotted them, so the labels are positional rather than an exact setting-for-setting match.",
          "points": [
            {
              "effort": "low",
              "cost": 1,
              "score": 2.1
            },
            {
              "effort": "medium",
              "cost": 2.52,
              "score": 14.3
            },
            {
              "effort": "high",
              "cost": 3.35,
              "score": 22.6
            },
            {
              "effort": "xhigh",
              "cost": 5.63,
              "score": 29.2
            },
            {
              "effort": "max",
              "cost": 11.14,
              "score": 37.3
            }
          ]
        }
      ]
    },
    {
      "id": "openai-terminal-bench-2-1-agentic-terminal",
      "suite": "Terminal-Bench 2.1",
      "task": "Agentic terminal tasks",
      "x_label": "API cost per task (USD, log scale)",
      "y_label": "Score (%)",
      "as_of": "2026-07-09",
      "publisher": "OpenAI",
      "source_kind": "vendor-reported",
      "source": "https://openai.com/index/gpt-5-6/",
      "confidence": "medium",
      "method": "Digitised from the interactive score-vs-cost chart in OpenAI's GPT-5.6 launch post: point positions were pulled from the chart's SVG and calibrated against the six values the page states in its table (Sol 88.8, GPT-5.5 85.6, Fable 5 83.1, Opus 4.8 78.9, Gemini 3.1 Pro 70.7 all reproduce to ±0.1), so treat scores as ±0.3 points and costs as ±$0.03. Effort labels are not in the chart markup; each model's points are ordered by cost and labelled low → max by position, which matches the page's statement that Sol ships with xhigh and max rungs. GPT-5.5 has four points. Models with a single point (Fable 5, Opus 4.8, Gemini 3.1 Pro) are not plotted — one point isn't a ladder.",
      "harness": "OpenAI's own run of Terminal-Bench 2.1. Cost is OpenAI's estimate, not a measurement: the page's footnote says latency and API cost were estimated by looking at production behaviour and simulating offline at regular API pricing, and that real-world results may vary substantially.",
      "caveat": "A lab publishing about its own models, on a chart it drew. Terminal-Bench is an accepted third-party benchmark, but this run, harness and cost model are OpenAI's and have not been reproduced. Two things to know: the page's summary table states Terra 87.4% and Luna 84.7%, higher than the top of their plotted curves (84.3 / 82.5) with no explanation; and OpenAI cut Luna's price 80% and Terra's 20% on Jul 30 and Sol's 20% on Aug 21, so the plotted costs are launch-day prices and now overstate what those rungs cost.",
      "levels": [
        "low",
        "medium",
        "high",
        "xhigh",
        "max"
      ],
      "series": [
        {
          "model_id": "gpt-5-6-sol",
          "label": "GPT-5.6 Sol",
          "color": "#9aa0ab",
          "level_note": "Labels are positional (ordered by cost), not read from the chart.",
          "points": [
            {
              "effort": "low",
              "cost": 0.66,
              "score": 76.4
            },
            {
              "effort": "medium",
              "cost": 0.89,
              "score": 81.8
            },
            {
              "effort": "high",
              "cost": 1.09,
              "score": 84.7
            },
            {
              "effort": "xhigh",
              "cost": 1.33,
              "score": 84.9
            },
            {
              "effort": "max",
              "cost": 1.72,
              "score": 88.8
            }
          ]
        },
        {
          "model_id": "gpt-5-6-terra",
          "label": "GPT-5.6 Terra",
          "color": "#e8a93a",
          "level_note": "Labels are positional (ordered by cost), not read from the chart.",
          "points": [
            {
              "effort": "low",
              "cost": 0.24,
              "score": 55.5
            },
            {
              "effort": "medium",
              "cost": 0.41,
              "score": 71.7
            },
            {
              "effort": "high",
              "cost": 0.63,
              "score": 76.6
            },
            {
              "effort": "xhigh",
              "cost": 0.73,
              "score": 79.6
            },
            {
              "effort": "max",
              "cost": 0.97,
              "score": 84.3
            }
          ]
        },
        {
          "model_id": "gpt-5-6-luna",
          "label": "GPT-5.6 Luna",
          "color": "#7fd9b4",
          "level_note": "Labels are positional (ordered by cost), not read from the chart.",
          "points": [
            {
              "effort": "low",
              "cost": 0.13,
              "score": 49.4
            },
            {
              "effort": "medium",
              "cost": 0.25,
              "score": 58.7
            },
            {
              "effort": "high",
              "cost": 0.44,
              "score": 73.9
            },
            {
              "effort": "xhigh",
              "cost": 0.59,
              "score": 74.6
            },
            {
              "effort": "max",
              "cost": 0.63,
              "score": 82.5
            }
          ]
        },
        {
          "model_id": "gpt-5-5",
          "label": "GPT-5.5",
          "color": "#3f7fe0",
          "level_note": "Four points on the chart; labelled low → xhigh by cost order.",
          "points": [
            {
              "effort": "low",
              "cost": 0.35,
              "score": 72.8
            },
            {
              "effort": "medium",
              "cost": 0.54,
              "score": 80
            },
            {
              "effort": "high",
              "cost": 0.91,
              "score": 83.8
            },
            {
              "effort": "xhigh",
              "cost": 1.33,
              "score": 85.6
            }
          ]
        }
      ]
    }
  ],
  "effort_ladder_note": "effort_ladders = published cost-vs-performance curves where a single model is run at several effort/reasoning settings, so you can see what extra spend actually buys. Each ladder carries its publisher, harness, method and confidence, and is shown with those attached — vendor-run benchmarks are directional, never a settled ranking. A ladder is only listed when the publisher has released the figure; nothing here is modelled or extrapolated from list prices."
}